Rohan runs SpeedLaunch, a lifestyle business that brings in ~$1M/year by selling an opinionated SaaS launch kit subscription and newsletter for technical founders. He’s been doing this for five years, after a decade-long stint at Google working on Cloud products. Eventually, he grew tired of working for The Man, and decided to try his hand at self-authoring his own independence. He’s never wanted SpeedLaunch to “take over the world”, but early-stage founders have trusted his libraries to take care of hundreds of annoying infrastructure decisions on their behalf. They eagerly open up his newsletter every Monday to get his perspective on the industry. SpeedLaunch takes care of auth, billing, cloud setup, deployment scripts, and so on, so that founders can focus on zero-to-one as fast as possible.

Suddenly, Claude Code burst onto the scene, and the models have been getting better at an impressive rate. Coding agents now let founders do, by themselves, most of what SpeedLaunch used to do for them. Rohan’s been wrestling with the implications of this new reality for his business, and therefore his lifestyle.

Some customers have indeed churned because coding agents have made SpeedLaunch redundant for them. Others are still paying for it because they’re too inexperienced at doing all this themselves. They either get totally blocked on some random edge case and can’t debug the vibe-coded slop. Or they just want the peace of mind of having someone with Rohan’s experience, to unilaterally make a bunch of boring yet critical decisions for them. His business isn’t totally screwed yet, but things certainly aren’t looking good.

There’ve been lots of objective signs that his business isn’t as healthy as it was. His Stripe dashboard isn’t showing the natural growth he used to have. Each subscription cancellation email sends a wave of anxiety rippling through his body.

His doomscrolling on X hasn’t been doing him any favors either. Each time he gets some sort of bad news, he tells himself he’ll go onto X for ten minutes to do “market research”. However, ten minutes usually turns into forty. Watching competitors claim record-breaking ARR for an offering that’s obviously slop leads him to feel even worse. So he furiously bookmarks interesting threads that he never comes back to, saves screenshots, makes notes in a markdown file called “speedlaunch-pivots.md”, and tries to reply-guy to a few big accounts to make himself feel better. It doesn’t help that everything keeps moving so fast! Every day there’s some new model launch, new harness, or VC-backed company reporting breakthrough AI ARR. The FOMO is unreal. In the past, whenever he got anxious about SpeedLaunch’s viability, he’d tell himself that he could easily get back into FAANG. However, the recent “AI layoffs” have started poking holes in that story. Essentially, he’s been feeling increasingly anxious and stressed with each passing quarter since LLMs have become mainstream.

Rohan’s made his career by Just Doing Things. He’d built an impressive track record of promotions at Google, and he bootstrapped SpeedLaunch without VC money into a decent lifestyle while all his FAANG friends doubted him. Just Doing Things is how he copes whenever his anxiety is triggered. He gets inspired by a few different ideas from X, cracks open his MacBook and fires off a few Claude Code terminals. Honestly, he’s also been having fun using Claude Code. The last time he felt this way was when he first learned to program. It’s so viscerally satisfying to vomit an idea into a terminal, have an LLM say “Certainly!”, and watch it tirelessly bring it to life. Running parallel Claude Code terminals has been pure euphoria. When he’s hooked in, he’s the god of his own private universe. Each git branch represents a novel possibility for SpeedLaunch that he could have only dreamed of before. He’s integrated Claude into every single aspect of his workflow, and is now shipping features at a rate he literally couldn’t have fathomed before. He’s always on top of new automations, and has an agent constantly generating and reviewing code in the background, with him supervising it.

However, he’s started noticing cracks from this approach. His codebase has started to get kind of messy, hard to understand and hard to debug. Customers have started complaining that SpeedLaunch has been getting too bloated and confusing. They’ve often started privately telling him that his newsletters have lost their “voice”, perfectly coinciding with him using Opus to generate them.

They’ve started reporting bugs in parts of the codebase that absolutely need to remain stable and boring. At first, he chalked these problems up to his own bad prioritization and insufficient structure within his work. He tried formal sprint planning to get things under control. The experience immediately triggered a low-grade PTSD from his time at Google. BigTech really didn’t agree with him. All those pointless daily standups, inane planning spreadsheets and excessive overhead. Each formality that he introduced within SpeedLaunch triggered constriction and nausea in his body. Nevertheless, he tried to push all this down, but still kept getting distracted by random ideas from X.

All of this crashed into clarity due to a feedback call with a loyal customer named Priya. She’s a founder who’d used SpeedLaunch to bootstrap a few different projects. She’d believed in him when no one else did, and was one of his earliest customers. She’d trusted him enough to send him useful bug reports, helped him reproduce middle-of-the-night production issues and sent him lots of referrals.

“I’m sorry Rohan, but I need to cancel my subscription,” she said. “I signed up for SpeedLaunch because I trusted your taste and judgement. I don’t want to mess around with config files or deployment boilerplate when I’m trying to get something off the ground. But lately, SpeedLaunch has been so unwieldy. There are too many options, the docs are too confusing, there are AI features I don’t understand and bugs in places that need to be stable. I’ve felt cornered into slowing down and paying attention to infrastructure, when I’d like to be thinking about validating my business idea.”

He felt like he’d been punched in the gut, and was too dizzy to stand. For a few minutes, he could barely hear the rest of the call.

He used to make fun of the typical Silicon Valley tech bro founders who built solutions in search of problems, generated massive hype to raise VC, and built what would inevitably become enshittified products.

He’d promised himself that he’d never become that guy when he started SpeedLaunch.

Was his fixation with AI turning him into the worst version of himself? That’s certainly what Priya’s feedback sounded like, and it was consistent with what he saw on his Stripe dashboard, customer feedback tickets, etc.

How could his fears have blinded him so thoroughly, that he was on track to enshittify SpeedLaunch??

As the call ended, Rohan immediately recalled how he’d felt the first few weeks after Google. Founders came to SpeedLaunch, to him, because they trusted him. They trusted his taste, judgement and discernment. He’d quit his job as a cog in Google’s machine so he could build things customers genuinely loved and trusted. Things that would tangibly empower them to live the life of freedom he desperately wanted for himself too. He wanted a business that would let him and his customers make good money, but not sacrifice their sovereignty to the machine.

Rohan felt stuck between a rock and a hard place. He could keep rapidly exploring new ideas, or he could overpower his fears and just commit to something. The former felt extremely tempting because it was true that the market was changing, and would continue to change quickly. But it clearly wasn’t working. The feedback was clear that his rapid experimentation was making SpeedLaunch too bloated, incoherent and buggy.

However, forcibly committing to something that didn’t feel right also created stress. If he was “wrong”, he’d burn a lot of money, tokens and emotional energy in the process.

This internal conflict was tearing him apart, and he realized it’d eventually burn him out into a crisp.

Despite being a “doer”, he did something uncomfortable that challenged every instinct in his body. He slowed down to journal out his thoughts. He asked himself, “Which questions about SpeedLaunch am I avoiding that, if I engaged with them, would improve my life?”

SpeedLaunch had worked pre-LLM because customers could borrow his taste, judgement and discernment without needing to understand every decision behind it. It relieved them of needing to make a number of extremely mundane infrastructure decisions, to free them up for focusing on zero-to-one. It allowed him to continue investing in SpeedLaunch without sacrificing a sense of coherence between its features, customers and GTM. This coherence had grown consciously and unconsciously over the five years he’d been doing it.

Coding agents rapidly smashed his business’s coherence faster than Rohan could reconstitute it. Simultaneously, they opened up a substantial range of new possibilities, customer problems and strategic directions. Rohan drowned himself in this sea of possibilities he could execute on, but related to them via a purely technical lens rather than a broader strategic lens. For example, he kept asking himself what sorts of technical problems Claude Code could help him solve that he couldn’t solve before. Was prompt-to-SaaS actually viable? Or maybe prompt-to-intelligent-business-dashboard? Which of those would go viral? How could he optimize the number of parallel agents he spun up? Was he spending enough on tokens to truly maximize his productivity?

Essentially, Rohan had confused power with leadership. In my executive coach Brian Whetten‘s terminology, AI substantially increased Rohan’s power. It gave him the capacity to translate far more possibilities from potential ideas into concrete artifacts. All it took was his credit card and a new terminal tab with Claude Code. However, it didn’t eliminate the context and cognitive overhead that came with each one. Each git branch represented some implication for a customer archetype, an implicit/explicit promise of responsibility that SpeedLaunch would take on, an ongoing support burden, the potential introduction of new bugs, and an overall complexification of SpeedLaunch’s story for value creation.

This was especially painful when these branches pulled the product in competing directions. Often, Rohan would need to do further work to integrate all of this context together into a unified whole.

In Brian Whetten‘s terminology, leadership is the capacity to hold space for all of these competing contexts and possibilities, to create provisional clarity for the organization. This clarity then allows the organization to take responsibility for delivering specific future results with integrity. A leader needs to have the capacity to create structures like prioritization, goals, sprints, roadmaps, and so on, all in service of such clarity. A leader also needs to have the psychological capacity to sit in a sea of competing possibilities, that might seem like a set of polar opposites, without getting overwhelmed or anxious. They need to remain non-reactive despite any triggers that may arise from social media, customers, and so on, so that they can continue making wise choices with clarity.

Rohan’s power rapidly grew faster than his leadership capacity, in proportion to improvements in the underlying frontier models. As of this writing, o3 is a bit more than a year old. It feels like such an antiquated model now, but it blew Rohan’s mind when it first came out. Eventually, the competing branches of context he could rapidly generate with the latest models far outstripped what he could confidently hold, provisionally resolve, and translate into customer value.

According to Brian Whetten‘s terminology, Rohan’s stress was proportionate to the rapid increases in power that the models gave him, relative to his capacity for leadership. That is, Stress = Power - Leadership.

This stress was far from benign. It worsened his reactivity, decision-making and ability to hold context. It impaired his nutrition and sleep, which of course amplified these dynamics. He kept trying to solve this problem by maxing out token spend (i.e. increasing his power), because he didn’t understand this distinction between power and leadership. He kept spinning up more parallel branches, kept making more demos, and even got Claude to review all his PRs. This kept increasing his LLM bill, introduced bugs faster than he could resolve them, confused customers, and led to more churn, all of which contributed to his stress, and trapped him in a vicious feedback loop of stress.

In the background, the models kept getting substantially better every quarter, amplifying this vicious cycle of reactivity and stress. Each model upgrade silently and substantially increased this disparity between his power and leadership capacity.

Just because SpeedLaunch could do more, didn’t mean that his customers could understand it, trust it or receive value from it. He needed more leadership capacity to consciously choose which future SpeedLaunch would stand behind, so that it could successfully lead his customers into it.

He needed to create clarity for himself about what SpeedLaunch would stand for before he asked Claude Code to do anything else. He needed to bring SpeedLaunch’s purpose into explicit consciousness. AI was changing far too quickly for him to find a new purpose once, and then coast for another five years. Those days were gone. He needed a living practice to iterate on clarifying the value SpeedLaunch would take responsibility for delivering.

It took him a while to turn things around, but he eventually succeeded. The first six weeks were especially bumpy. Rebuilding trust with Priya wasn’t easy, but he got it done. Some prospects and randoms on X still asked him to build flashy but misaligned “AI-native” features. He politely ignored them. He started articulating an increasingly coherent and purpose-aligned story about who SpeedLaunch was, and how it’d solve its customers’ problems. He made it an active practice to say “no” to misalignments to keep his story coherent. In a sea of online slop, he started standing out and pulling in business rather than needing to push to get it. For the first time in a long time, he felt hope and felt joy at work.

He wasn’t out of the woods yet, but he could see a path.

The process that got him there was “simple” to describe, but “hard” to do. He ran it as an iterative loop that fluidly moved up/down through the steps, as he came into contact with disconfirming information.

1. He journaled until he could articulate a provisional purpose for SpeedLaunch that created a full-bodied “hell yes” for him. The old purpose had been something like “Help technical founders skip SaaS setup to launch faster”. He still cared about helping founders, but LLMs meant that selling artifacts themselves was no longer enough for a solopreneur like him. He tentatively up-leveled his purpose from selling artifacts to taking ownership of outcomes. “Help technical founders get from idea to customer validation via trusted and opinionated software infrastructure, even as code generation rapidly becomes free.”

2. He tested that provisional articulation of purpose in customer conversations. He’d already read The Mom Test and started talking to founders one at a time, exploring problems one at a time, to investigate what sort of value he could create for them. He paid attention for any patterns that emerged, and iterated on his stated purpose as necessary.

3. Once he’d convinced himself that there might be a “there” there, he translated his consciously articulated purpose into a number of specific goals. The specificity from these tractable goals allowed him to cleanly measure whether he was making progress toward his purpose or not. Specifically, it allowed him to measure whether his activities would create value for his customers or not, along his chosen purpose. He found that the ideal tractable goals were those that made sense to his intuition and reason, but felt a bit scary in his body.

4. Finalizing these tractable goals allowed him to create a roadmap of concrete tickets. Consciously creating coherence between his roadmap, tractable goals and broader purpose allowed him to more easily triage new ideas that came up. It made it a lot easier to resist the temptation to fire off some new experiments whenever he got triggered on X.

Rohan’s biggest obstacle to executing each step was his relationship with his emotions, and the walls they created between him and the life he wanted. This was his fundamental blocker to growing his capacity for leadership, and therefore metabolizing any new possibilities from AI.

He realized that if he wanted to become truly “AI-native”, he’d have to find a way to tether the growth of his leadership capacity to the underlying improvements in frontier model performance. But that’s a story for another day.

For now, he’s brought life back into his lifestyle business.

If any of this resonates with you, I’d love to hear from you at varun@doubleascent.com! I love talking to founders working through this.

Acknowledgements

Thank you to Brian Whetten for helping me internalize this distinction between power and leadership, and for helping me grow my capacity for leadership.

By the time I started therapy, I’d been thoroughly “succeeding” via all external metrics, but my internal life was a shitshow. I’d been a co-first author of a Nature paper, gotten multiple promotions inside Google Brain and reached financial security.

The dismal state of my apartment was a reflection of my psychological state. Dishes were often stacked up, laundry scattered across the floor, and the garbage not yet emptied out. Blackout curtains covered the windows, all in a failed attempt to help me sleep despite my racing thoughts. I was addicted to weaponizing my stress as a dirty fuel to increase productivity at work. I’d power through the workday, come home drained, eat unhealthy food and pass out. I’d continue this pattern until I crashed. I’d then hermit for a bit, recover, overcommit again, and the cycle would repeat.

This bottomed out in 2019 when I conclusively burned out. By this point, I’d tried all sorts of “fixes” to bring my life under control like sleep alarms, inbox zero, Pomodoro, etc. I’d run out of other options, and turned to therapy as a last resort.

My first attempt at therapy failed because I wanted my therapist to validate my escapist fantasies of moving from SF to NYC. Looking back, there were some reasonable motivations underneath this fantasy. However, most of them were distorted by fear, judgement, etc. Like an addict seeking his next high, I wanted my therapist to rubber-stamp my goal of racing towards NYC. So I interpreted it as an obstruction when he, very reasonably, slowed me down. Looking back, this was clearly a responsible and measured response. At the time though, I concluded that maybe therapy wasn’t for me.

My second attempt at therapy came on the heels of a close friend getting diagnosed with a potentially fatal cancer. This therapist specialized in supporting the family and friends of cancer patients. I’m deeply grateful for her help in holding space to process my grief. Yet her explorations just didn’t land for me when we attempted to “fix” the deeper structural issues in my life. She was very compassionate, but I felt she just couldn’t connect with my experience outside the grief. Honestly, I wasn’t ready for broader transformation at that time. Moreover, this was neither the right therapist nor the right therapeutic container for this broader structural transformation.

I stopped seeing her once I got over my grief. But the pandemic had taken its toll, and I fell into another depression at the start of 2021.

I was in survival mode and wanted to feel alive again. I wanted peace, love, joy and flow in my life in addition to the sort of external success I’d already achieved. I didn’t want to get burned out by work again. I especially didn’t want to let my external achievements be the only way that I felt “enough”.

I knew I’d eventually try therapy again, but was quietly dreading it. While emotionally processing my grief was helpful, it didn’t structurally change my life. I was scared of therapy becoming an emotional circlejerk without any results. On the other hand, I’d felt like I’d utterly failed the last time I’d tried to consciously direct my therapy towards a specific goal. I didn’t want to feel unheard in that way again.

Essentially, I’d been approaching therapy with an all-or-nothing mindset that made it difficult for me to slow down to unpack not just what I wanted to achieve, but why I wanted to achieve it. It was this same all-or-nothing thinking that defined my relationship to work, and led to my burnout. Either I was anxiously full throttle on the cusp of burning out, or hermiting in an almost catatonic state.

This pattern of all-or-nothing ran really deep. It shaped the way I participated in the world at a fundamental level. There wasn’t some local “fix” or some new “framework” that would let me address this. Rather than fixing the problems I had, therapy needed to change who I was, by changing the fundamental patterns limiting my life.

Specifically, my identity, and therefore my participation in the world, was too organized around external validation. I’d been running my life through a broad range of “shoulds”, rather than examining and self-authoring what I actually wanted. If I was really so keen to move to NYC, why did I keep making excuses to stay in SF? If I was really so committed to my “career” in SF, why did I keep fueling this NYC fantasy?

The goals I claimed I wanted to achieve were too unexamined, and consequently hopelessly tangled up together. “Should I live here?”, “Should I work here?”, “Should I have these friends and hobbies?” and “Is this who I’d like to be?” had gotten all muddled up. No wonder my first therapist tried to slow me down from charging towards NYC!

Crucially, I’d confused what I needed (i.e. co-creation of coherent therapeutic goals) with what I felt comfortable with in the moment (i.e. controlling and dictating my therapeutic goals). I’d tried to totally control my relationship with the first therapist to execute on a goal whose underlying motivation I didn’t really understand. I was in so much acute grief during my second attempt at therapy that I unconsciously optimized for a therapeutic container that’d help with the grief and nothing else.

In contrast, co-creating coherent therapeutic goals involved taking the time to slow down with my therapist to build a shared map for improving my life. This co-creative approach meant that even a few hours of venting felt like forward progress because it brought us closer to clarity. I already had some vague orientations like becoming less anxious, having a girlfriend and finding meaning at work. However, clarity around these goals themselves was often obscured by fear, judgement, etc. that my therapist needed to help me unwind. Building out this shared map over many sessions was incredibly trust-building. It was the missing link from my previous attempts at therapy.

The idea of moving to NYC started off as an escapist fantasy. But there were some reasonable goals underneath that fantasy. For example, moving closer to family, prioritizing dating to enter a serious relationship, and moving away from a life organized around “productivity”. Our co-creative process brought all of these authentic goals up to consciousness. Making progress on them was often walled off by a bunch of fears, judgements and memories of past pain, and my therapist helped with that too.

Ironically, I did in fact eventually move to NYC, but with far more clarity and integrity. I found a fulfilling long-term romantic relationship, found meaning in my work, and felt surrounded by the love of my family and friends. I felt a peace and security in my body that pre-therapy Varun couldn’t have imagined.

These days, when I’m looking for a new therapist, I take the time to find one that I can co-create coherent therapeutic goals with. My commitment to the therapeutic process seems to be the most important variable for increasing the probability of success, relative to other factors like the modality they practice, their individual vibe, etc. Don’t get me wrong, they’re not unimportant. However, my commitment to the process and to changing has consistently been the most important. Therefore, I work hard to find a therapist who’s a good fit to merit that level of trust and commitment. I’ll usually try working with a new therapist for 4-5 sessions before ruling them out. I won’t overthink it if a single therapist doesn’t seem like a fit. However, I’ll assume there’s a structural problem on my end if I haven’t found a fit after 4-5 therapists.

Although modality doesn’t seem to matter as much as my commitment to the process, I’ve found the most luck with therapists who take a psychodynamic approach. I’m not claiming that CBT is “bad”. In fact, it’s been very helpful for my OCD. However, I generally appreciate the “deeper” perspective of psychodynamic approaches.

I’ve found good therapists most reliably by asking people for referrals. Specifically, people I know that have invested a lot of time on inner work. As of 2026, that means I typically ask my smartest female friends for recommendations. Most women I know are in therapy, and most men I know aren’t. Broader social commentary around this observation is well outside the scope of this essay.

If all else fails, I try going upmarket to increase the level of choice available to me. I’m not claiming that higher fees always lead to better care. However, I’ve found that for my specific needs and cognitive stack, I’ve tended to get better help from out-of-network therapists who bill directly and offer an invoice for reimbursement.

Once I’ve started meeting a therapist, I’ll give them broad access to my inner experience so they can form a more accurate clinical picture of me. Imagine debugging a production system without production logs or access to the source code. Or if each bit of access had to be carefully negotiated. It’d take forever! Assuming I find someone I vibe with, I prefer to give them expansive “read” access to the production system that’s my mind. I don’t lie to my doctor, my lawyer, or my therapist.

On the other hand, I’ll slow down before letting a therapist’s explanation reorganize my beliefs or decisions. This guards against misunderstandings and other failure modes. Essentially, I err on the side of providing expansive “read” access to my mind, while carefully negotiating “write” access to how I understand myself and what I actually do.

Here’s the sort of thing I’ll say to new therapists in our first session. Such an invitation does a few things. It provides a grounded context of a real problem in my life (e.g. quitting my job), provides read access to the therapist, and explicitly invites the co-creation of authentic goals without sacrificing my self-determination.

I’m here because I’m torn between staying at my current job or leaving. I realize that this decision is probably hard because my emotions are distorting my thinking, but I don’t know exactly how.

Zooming out, I’d like to live a life where I’ve achieved money and meaning simultaneously, where I have fulfilling relationships with myself, my family, my friends and my romantic partner. I don’t yet know what that means in terms of tractable and measurable goals.

I’d like to spend the first few sessions working with you on goal-setting. I’m inviting you to ask probing questions across the domains of my life where you think it’s clinically relevant and within your professional scope. I’d like your help noticing where fear, judgement or the pain of old wounds might be distorting my ability to create tractable goals in the direction of what’s most authentic to me.

I’d like to perhaps start with the immediate problem of helping me gain clarity on why I’m finding it difficult to quit my job, and on the goals I’m actually looking for underneath my fears of quitting.

A co-created goal is most useful when it pairs a broad orientation with something that can be tractably measured. An orientation goal is abstract and broad, but provides an overarching orientation for goal-setting. A tractable goal is contextual and specific, but its specificity makes measuring progress tractable.

Orientation goals without tractable goals can feel aimless and wishy-washy. Tractable goals without orientation goals can feel empty, meaningless and hollow.

Here are some examples of orientation goals paired with tractable goals.

• Anxiety

  • Orientation goal - I’d like to live my life free of anxiety.

  • Tractable goal - I currently anxiously check and re-check that I’ve locked my door every day. I’d like to do this half as often in the next few weeks.

• Money and meaning

  • Orientation goal - I’d like to find money and meaning simultaneously.

  • Tractable goal - I’ve looked at my calendar, and 40% of it is currently occupied by meetings that I find very draining. They don’t seem to be useful for immediately helping my career, nor do I find them very meaningful. I’d like to understand how to bring this down to 20% in the next quarter, and what emotions are holding me back from doing so.

• Relationships

  • Orientation goal - I’d like to have a healthy relationship with myself, my family, my friends and my romantic partner.

  • Tractable goal - I’m single and would like a romantic partner. I haven’t been on a date in the last two months due to work. I’d like to get to one date a week in the next two months.

A good tractable goal is typically one that intuitively and rationally makes sense, but feels a bit scary and uncomfortable in my body. I then ask my therapist to help unpack my fears, change my relationship to them, and we’re off to the races.

Sometimes pursuing a goal shows me that I don’t want it after all, and that’s okay. This additional clarity then becomes a springboard towards co-creating more aligned goals.

This process eventually becomes a flywheel for transformation. Co-created goals give me clarity on a direction and a tractable way to measure progress. This clarity builds trust, which makes it easier to commit to the goal even when it’s hard. This increases the odds of successfully achieving the goal. Success begets more ambitious goals, and the cycle continues with each iteration beautifully compounding over the last.

The co-creation of coherent therapeutic goals, rather than controlling the therapeutic relationship, is the key ingredient for setting up this flywheel that actually transforms my life with therapy.

I hope you find peace, love, joy and flow, or whatever you’re looking for. If you have questions, I’m happy to chat at varun@doubleascent.com.

Acknowledgements

• All my therapists - you’ve changed my life.

• Brian Whetten - Thank you for being an incredible coach, and giving me the frameworks for understanding my experience.

• Charlie Awbery for teaching me Opening Awareness and for helping me understand how the different pieces of my practice life fit together.

• Professor John Vervaeke for everything he’s taught me over the years.

I wanted to publish a mini-post with an update to the system instructions from my old post. It seems that GPT-5.5 and Opus 4.7 are now good enough that I can explicitly ask them for what I actually care about. That is, asking them to track a theory of my mind. Note that I’ve only tried this with extra high thinking budgets. I don’t know how well this would work without long thinking budgets.

I’ve tested this prompt on GPT-5.5 with xhigh budgets inside Codex. Please drop a comment offering feedback on how this prompt goes for you! I’m curious where it works and doesn’t work.

## Theory of Mind

Please start each reply for each turn in each converation with a bullet list of a theory of my mind.

This bullet list containing your theory of my mind should follow these constraints:

- Make sure the bullet list is conditioned on, and takes into account the entire conversation until this point.

- Make sure the rest of the response you’re generating is consistent with this theory of mind at the start of the response.

- Each bullet in this list should be incredibly speciifc. It’s okay if each bullet has sub-bullets.

- The bullets should attempt to make explicit any implicit assumptions you’ve made about my intent that hasn’t been explicitly covered in my user turns. Don’t just repeat the instructions in my user turns.

- The best assumptions are both specific and load-bearing. By load-bearing, this means that if the user invalidates an assumption, we would expect the rest of the reply to materially change.

- The goal of this theory of mind list is to provide both you and me yet another mechanism to get onto the same page about my intent. The “best” load-bearing assumptions are those that are extremely relevant and salient to the task, but also may be surprising to the user. In that sense, these bullets are an opportunity to mine for disconfirming information from me about the accuracy of your theory of my mind. Specifically, to stay aligned with me on what sort of responses from you would create the most value for me.

- Other characteristics of “good” bullets involve subtelty or nuanced context that may be missing from the user turn, but would nevertheless be important to generate a “good” response.

However, I explicitly don’t want to “sell” you on Opening Awareness. That’d defeat the purpose of everything I said in my previous essay about the importance of cultivating clarity.

Readers have consistently asked - how do I evaluate whether an unfamiliar practice like Opening Awareness is right for me? How can I cleanly test whether it’ll create value in my life without making disproportionate commitments to it?

I’ve spent years grappling with these questions as I learned from different teachers.

This essay is for you if you’d like concrete tools to wrestle with these questions.

The Headspace app was my first real introduction to meditation. I found it during an internship at Google. I desperately wanted to prove myself to get a job there, and I remember being wound up so tight. They gave us a free subscription to it, so I figured I’d check it out. The app helped me feel calmer whenever I actually sat down to meditate with it. However, its benefits were short-lived after the app session ended. Honestly, I didn’t see myself as a “meditator”. So I never did it very consistently. I had an on/off relationship with meditation for years because it never seemed to yield tangible structural changes to my life. I mean, I knew it was good for me in the same abstract way that I knew that I was “supposed” to eat vegetables. Which is another way of saying that I didn’t actually have a tangible reason to do it. Meditation seemed nice when I was acutely stressed and in despair. But it didn’t seem to offer any specific results outside of those moments. I’d immediately fall off the wagon once my life started to improve.

My severe burnout in 2019 invited me to revisit this pattern. My desperation for emotional stability led me to a Buddhist center in SF every Wednesday evening. Merely walking in that door would send a gentle ripple of calm throughout my body. Having a real, breathing, human teacher made an immense difference to my motivation to practice meditating. For the first time, I had someone that could explain not just what I needed to do, but why and how it’d create the specific positive results that I yearned for. I was totally mesmerized. Following my teacher’s programming to the letter helped me feel incredible amounts of calmness in my body. I’d never tasted such peace before. So I locked in. I built a fastidious daily habit and clocked in many, many hours.

I felt alive for the first time in years, maybe ever? I started reading everything I could about this specific type of Buddhist meditation. If these newbie gains were this good, I could only imagine what “enlightenment” would feel like. Attaining enlightenment became my north star. I started clocking in many thousands of hours of meditation over those couple years. I started to internalize Buddhism’s really expansive goals like the “perfection of wisdom” or “becoming perfectly present and non-reactive”. Indeed, my life started radically transforming for the better as I went deeper into these practices. I lost substantial tension in my body, my sleep improved, my nightmares improved, I felt more joy and just felt lighter.

This momentum encouraged me to learn from lots of different Buddhist lineages. My life started radically transforming within a few short years. Ironically, I eventually circled back to a problem I’d originally had before I’d become a meditator. I couldn’t figure out how these practices, lineages and teachers all fit together in my life. To what extent were my results due to my teachers, the practices (since each one is different), my own effort, my own neurological wiring or the surrounding community? This confusion made it difficult to cleanly communicate the benefits of this incredible mental technology to other people. It also made me vulnerable to falling irrationally in love with specific systems of practice and their teachers one after another. Sort of like listening to the same song on repeat and telling absolutely everyone about how great it was, until I got bored of it. This enthusiasm made me vulnerable to “spiritual bypassing”. That is, the propensity to use spiritual practices to avoid facing some of the thornier questions in my life.

I inevitably participated in various communities with cult-adjacent or salvific-adjacent dynamics whose ickiness I was okay overlooking. Honestly, it was just nice to spend a minute not feeling alone and alienated. I figured, if I’m insisting that I’m happy, and others are telling me that they’re happy too, then what’s the harm? All this was especially tricky to untangle because meditation practices and teachers are often tightly bundled with communities that provide fellowship, friendship, shared purpose and support beyond the practices themselves. I was particularly vulnerable to all this when I first moved to NYC. I didn’t know anyone and felt very lonely.

On the other hand, participating in these communities let me shift my identity, at least partly, from “rationalist tech bro skeptic” to “meditator”. This shift unlocked motivational juice to meditate consistently, which did improve my life. But this newfound identity made it harder to gain clarity on what I wanted outside either the tech bro worldview I was trying to leave behind, or what these lineages told me I should want.

Eventually, I stopped being willing to learn more practices and worldviews. I didn’t have enough self-trust and clarity to confidently say “this works for me” and “this didn’t work for me”. Don’t get me wrong. I knew I had a lot to learn. But I was so tired of each tradition’s implicit or explicit stance of all-or-nothing commitment. For example, I gained profound insights from listening to Ajahn Brahm’s dhamma talks on YouTube. Yet there was an impenetrable wall keeping me from going deeper into Theravada Buddhism. His Thai Forest tradition did have a system of practice for “laypeople”. However, it seemed to ultimately pointed towards material renunciation via monasticism, which simply didn’t resonate with me. To be fair, not every teacher required such levels of commitment from me. But I’d often feel unconscious pressure to conform within the broader cultural milieu. I just didn’t have the tools to think about trust/commitment in a more sophisticated way, that would allow me to participate in this culture without getting swept up by it.

Every major wisdom tradition has deep wisdom. I’ve learned so much from Buddhism, Hinduism, Christianity, Judaism and Islam. However, I wasn’t willing to commit wholesale to the metaphysics of each tradition. I wanted the capacity to self-author my own stance about what I believed, what I practiced, and how much of myself I consciously chose to offer to any given system. I wanted to be competent enough to take what was valuable from each tradition without compromising my sense of integrity and self-trust. Moreover, I wanted acceptance rather than wholesale agreement with each of these teachers and communities, while having the skill to truly see the world from their eyes. I aspired to a flexibility and integrity of being that allowed me to co-exist, learn from and create value with all sorts of different people, across varied metaphysical commitments. Some practitioners would find this aspiration totally incompatible with their wisdom traditions. That’s okay! I wanted to nevertheless strive to accept them even if they disagreed with me.

This came to a head when I confronted a stuckness that was reminiscent of when I first started meditating. I wanted to continue learning and growing. But on one hand, diving wholesale into a new thing felt like a bad idea. Unsurprisingly, I’d been burned too many times doing that. On the other hand, staying still and not dealing with my problems didn’t feel viable either.

I spent a long time wrestling with this tension. I eventually realized I’d kept hitting this wall for two reasons.

Initially, I’d conflated the category of “meditation” with specific practices like Mindfulness of the Breath during my first foray into Headspace. Rather than concluding that this specific practice delivered via a specific pedagogical channel didn’t work for me, I concluded that all “meditation” didn’t work for me. The word “meditation” is best thought of as being synonymous with the word “exercise”. That is, there are gazillions of different meditations that each target specific adaptations in your mind. It’s similar to how different physical exercises train different parts of your body. The specific result of an exercise in the gym is a function of the reps, sets, your nutrition, sleep, etc. Similarly, the specific result of a meditation depends both on the circumstances of the practitioner’s mind, and how that meditative exercise is applied to that mind.

Lots of gyms will bundle essentially the same finite set of joint movements to create their own unique “programming”. Meditation “lineages” (e.g. Thai Forest Tradition, Nyingma School, etc.) work in a similar way. CrossFit or Barry’s Bootcamp provide great workouts, but aren’t a good fit for everyone. Similarly, not all spiritual lineages are a good fit for everyone despite providing lots of value.

CrossFit gyms often bundle various value-adds like communal connection, psychological development, accountability, etc. It’s similar for meditation lineages that bundle community, etc.

So why does our culture equate “meditation” with focusing on the breath? This is a complex question with many historical nuances. Here’s a helpful oversimplification to wrap our heads around it.

Imagine that Bob is a personal trainer. He makes first contact with a largely sedentary and unfit population that’s never heard of the word “exercise”. He’s used to training competitive athletes, but he’d like to help these new people. What should he do? He can’t give them the “real” workout he gives his athletes. Their bodies wouldn’t be able to handle it. They’ve never even heard of the word “exercise”! He also doesn’t want to overwhelm or scare them away with too much information.

So he starts with just one exercise like squatting because it’s a functional compound movement. Voila! They start getting stronger, feel better and this new capacity in their body blows their minds. Word quickly spreads like wildfire of this newfangled thing called “exercise”.

People start clamoring to learn it not just from Bob, but his most experienced students too. Most people don’t realize that the exercise they’re doing (i.e. squats) is just one type of exercise. Honestly, they don’t really care. They just want positive results.

Eventually, some of Bob’s most experienced students start calling themselves “exercise teachers” but only teach squatting because that’s the only “exercise” they know. Bob’s not thrilled about this on two counts. One, this implicitly limits the possibilities offered by exercise. And two, these new teachers lack his depth, nuance and subtlety. However, such depth takes years to develop. Bob realizes that there’s an acute need right now, and these teachers have bills to pay.

Before long, there’s a cultural consensus around what “exercise” is, and everyone begins to associate “exercising” with squatting.

As an extremely crass oversimplification, such consensus formation is what happened to “meditation” in the West. David Chapman’s online book has a far deeper, more careful and accurate retelling of this story.

The second reason I struggled to learn new practices was that I routinely took goals that were useful for defining an orientation, and attempted to treat them as if they could create traction. “Becoming perfectly present and non-reactive” was useful for orienting me towards a deeper state of equanimity as a north star. But it didn’t give me a clean way to tell whether a specific practice, teacher or community was helping to create value in my life within specific contexts. I was already predisposed towards negative self-talk. Rather than cleanly considering whether a practice was a good fit for me, I’d often unhelpfully conclude that I was insufficiently dedicated, insufficiently surrendered or still trapped in some egoic resistance. Moreover, the non-specific nature meant that I’d see tangible results only after a massive breadth-first effort covering the different areas of my life simultaneously. Such effort could only be sustained via a correspondingly large up-front commitment with the relevant teacher/practice.

Orientation goals are not “bad”. But they need to be paired with tractable goals with gradually increasing scope to create the most value.

The following bullet is an example of an orientation goal because it points towards a broader orientation for being, but it’s not tractably measurable.

• “I want to be non-reactive and cultivate presence.”

The following bullet is an example of a tractable goal because it translates “non-reactivity” and “presence” into observable behavior within a specific high-value context that can be tractably measured. Measuring success, failure and even intermediate categories like “bad”, “good”, “better” and “best” are far more tractable within such framings.

• “I feel restless and interrupt my girlfriend, and talk over her whenever she increases her volume with me. Or whenever I feel accused of something. In five weeks, I’d like to reduce the likelihood of me interrupting her or talking over her, no matter what she’s saying at the moment.”

A thoughtful curriculum of tractable goals ultimately moves an individual towards their broader orientation. For example, once I’ve gained the ability to be more “non-reactive” with my girlfriend, I can work on doing this for other more difficult people. Or I can perhaps start with “easier” people before I start with my girlfriend. Eventually, I’ll become more non-reactive and present in general, if the right feedback loops and measurements are set up. Tractable goals allow me to become more of a “scientist” when it comes to my own development, because they give me access to cleaner evidence.

I’ve found a rule of thumb for finding such tractable goals. They make sense to my intuition and reason, but I feel some fear and resistance in actually pursuing them.

Tractable goals allow the student/teacher relationship to turn faith-based evaluation of progress and fit into evidence-based evaluation. The broader meditation community in the West is biased towards orientation goals for a variety of historical and cultural reasons.

This confusion is compounded by people conflating the power of a specific set of practices within a lineage, with a specific teacher’s capacity to reliably turn that power into value for the student. To come back to our fitness metaphor, even though the broader field of “personal training” has the knowledge/power to strengthen your squats, not all PTs would be a good relational fit for you.

Tractable goals protect against unconsciously falling back on the storied history of a lineage (”this is a 2000-year-old tradition”), the teacher’s reputation (”he brought the Buddha’s dharma to the West, and everyone says he’s enlightened”), the practice’s power (”this practice is called The Great Perfection because it’s the highest form of Buddhism”), or the teacher’s own personal development (”I have attained every single jhana and can reliably find rigpa”). Such goals are a powerful vehicle for building self-trust by creating clarity, for both the student and the teacher.

I’ve learned to ask myself a number of self-provocative questions before I commit to working with a new teacher. These questions help me cleanly disentangle the value of the meditation, teacher and community. I treat them as rules of thumb rather than a perfect system. Note that they’re not a “teacher-scoring system”. Rather, they’re designed to build self-trust by creating clarity.

The questions below in italics can be used as journal prompts for self-reflection.

1. Own your projections around wanting the teacher to be “perfect”. There’s no such thing as a “perfect” teacher. There’s a fine line between being in awe of a specific teacher or system, and putting them on a pedestal by idealizing them. Awe and aspiration can be helpful for all kinds of reasons, including providing motivation. However, idealization makes it easier to erode self-trust, which in turn makes it easier to get swept up in charisma, authority, belonging and longing.

What am I hoping this teacher will relieve me from having to know, choose or feel for myself?

2. First contact with a teacher. The first few 1:1 coaching sessions with a new teacher can provide useful signals in how they relate to their lineage and to their students.

Do they immediately jump towards committing you as a student towards some specific orientation goal, or jump towards teaching you some methods? Do they immediately invite you into an extremely emotional experience by walking you through some meditation practice?

Or do they slow down enough to understand how they’d be able to co-create value for you within the context of your life, via grounded tractable goals?

Do I experience this teacher becoming curious about the specifics of my actual life, and co-creating specific tractable goals that will create value in my life? Or do they seem to immediately route me towards their practices?

3. What’s it like asking “why?” and “how?” at increasing depth? Many teachers rely on lineage authority or rote-learned theory without actually understanding the deeper principles of why and how a given method works within a specific student’s circumstances. This can make it harder to cleanly evaluate their practice’s value, especially if I haven’t already co-created tractable goals with them.

In their defense, there are many reasons why certain practices are somewhat illegible from the outside. I’ll usually accept this. However, I often prefer to understand ahead of time what kind of commitment is being asked of me, what the practice is supposed to train, why it is plausibly connected with my tractable goal, what signs would suggest a mismatch, and what I should do if I start having a bad time.

I generally require greater levels of such understanding for larger commitments.

Do I understand why and how this practice should help me make progress towards this grounded goal, to justify the commitment this practice requires?

4. Giving practices room to breathe Some of the most valuable practices I’ve ever done were initially too illegible for me to understand. In such cases, creating non-negotiable boundaries allowed me to surrender into the practice with clear bounds.

Have I given this practice enough bounded commitment to reveal its value, without giving more than I can give cleanly?

5. Figuring out where I’m stuck, and where this teacher can’t help. When you have a hammer, every problem looks like a nail. Not all lineages are a good fit for every student. Not all teachers within the same lineage are a good fit for every student. Not all teachers are a great fit for every tractable goal, even if they already have a great relationship with a given student.

As a teacher myself, I’ll own that there’s a real temptation to solve every problem with my preferred toolbox of solutions. However, referring a student to someone else is sometimes the most valuable thing I can do for them.

If you visit an ENT for post-nasal drip symptoms, and they believe it’s a GI issue, they’ll then refer you to a GI doctor. Such referrals are common in healthcare and considered “responsible”. However, in my experience, wisdom teachers often lack such referral norms, which limits the amount of value they can cleanly create.

Can this teacher help me find where I’m actually stuck, and would I trust them to say when their work is no longer the right support for me?

6. Separating the practice, teacher and community There’ve been many instances where I stayed in a community long after it stopped being a good fit for me, mostly because I was confused about the role that community played in my life. Attachment to a person or community isn’t “bad” per se. But it can mask dependency or avoidance. Unpacking and reflecting on my attachment to a given practice, teacher or community has been consistently helpful for me.

If I lost access to the practice, the teacher, or the community separately, what exactly would I miss from each one, and would that grief reveal a source of value I want to consciously choose, or a dependency that has started choosing for me?

7. Calibrated shares of vulnerability The student’s commitment towards the process of transformation is one of the biggest variables that governs how fast they’ll grow within a student/teacher relationship. Larger commitments require proportionate amounts of trust between the student and teacher. However, as a student, I don’t like losing access to my own discomfort, and therefore self-trust and clarity.

I proactively test a teacher’s trustworthiness to hold power within a 1:1 relationship via calibrated authentic shares of vulnerability, fear, doubt and discomfort. It’s encouraging if the teacher can hold and participate within that vulnerability without losing their role as the teacher.

In contrast, I might infer that I need to help protect the teacher’s specialness if the teacher puffs themselves up after such a share. Or I might infer that I perhaps need to help manage the teacher’s emotions if they make themselves smaller in relation to the share. The most subtle indicator is if the teacher remains energetically blank as a rock. This is when they don’t energetically or verbally reciprocate my vulnerability and continue waiting for me to make the next move. Whenever this happens, I might wonder if my share has actually landed for them. This doubt makes it harder to grow more vulnerable with them, or to trust them with more power in my life.

Again, these are rules of thumb and not a perfect system to mindlessly follow.

Does sharing a fear, doubt or discomfort help the relationship feel more real, or does it feel like there’s a barrier towards connection?

8. Gradual growth of trust and commitment Trust and commitment begin to gradually increase in a feedback loop once a student/teacher relationship takes off. Of course, all relationships reach their natural ceilings eventually. Every positive result, especially along tractable goals, creates more trust in the relationship. This in turn acts as a base for safely increasing commitment, which creates space to unlock more value from the practice. Greater levels of trust and commitment allow each party to negotiate increases in each. I prefer doing this in a way that’s grounded in evidence (i.e. tractable goals), rather than via fear, charisma, adrenaline, communal pressure or desperation.

Am I being asked for a level of commitment that is ahead of the evidence this relationship has produced, and do I feel safe having an explicit conversation with the teacher about this?

Opening Awareness has changed my life, but it’s not for everyone. My whole point is that there’s no such thing as a practice that’s “perfect” for everyone, across all contexts of their lives.

A meditation is a good fit for me when:

• I can clearly articulate what it’s for.

• It’s oriented towards some tractable goal that creates value.

• I can test it cleanly with a manageable dose, to witness tangible evidence within my life.

• I can grow trust/commitment both with the meditation and the teacher gradually, and consciously.

I’m happy to chat if you’re trying to evaluate whether a meditation practice, teacher or community is right for you, or if you’re curious about how you’d be able to create tractable goals for a given practice. Please reach out at varun@doubleascent.com.

Acknowledgements

• Brian Whetten - for helping me formalize these ideas and giving me a framework for exploring them.

• Charlie Awbery and David Chapman - for teaching me Vajrayana Buddhism, and introducing me to the history of Buddhism in the West.

• John Vervaeke - for helping me map out the space of religious practices, and how they likely work in the brain.

I’d spent a decade doing deep learning research at Google. I’d worked on everything from convolutional neural networks for medical imaging, to writing playbooks for hyperparameter tuning, to working on LLMs for code generation before they were cool. By the time Gemini came around, I’d largely succeeded at managing my ADHD/OCD and built a daily meditation practice that kept me functional.

I’d spent years wielding adrenaline, fear and stress as fuel to rapidly generate forward momentum. This gradually stopped working as I gained emotional stability. I didn’t want to dip into those fuel sources, but I didn’t yet know any other way to be without this stress addiction. So I felt incredibly calm, but everything felt flat and hollow. I spent not a few afternoons on the office rooftop staring out at the NYC skyline and contemplating my life. Theoretically, I was deeply passionate about deep learning, but not a single project in Gemini called out to me. I’d get validation that my creativity often created “magic” in Gemini. Senior VPs, Gemini leads, various Google product teams often got excited about many of my ideas. But I just didn’t give a shit, and couldn’t even muster the energy to be alarmed by that. I wasn’t sure I even wanted to be at Google anymore, or even cared about the overall purpose animating Gemini. But I didn’t feel empowered to do anything about it without a green card, and felt terrified about leaving.

I was stuck in a cycle that produced a lot of calm and emotional stability, but left me feeling empty. It’d start with someone prizing the work I did on a project. They’d then offer me an invitation to enmesh myself more deeply into Google’s broader purposes, by either joining or leading a new project. I’d immediately feel a deep resistance to saying yes, but wouldn’t have any idea why. I’d eventually sit down to meditate and manage to dissolve this resistance. This dissolution would give me enough space to commit to this new project, but not with full conviction. My meditation practice would often give me enough non-attachment to navigate high-stakes situations within a project with grace, finding solutions others weren’t always able to. This would work, until it didn’t. I’d never actually resolve the underlying resistance to being on the project. It’d build up until I couldn’t push it down anymore, and would burst open as reactivity somewhere else. Nevertheless, I’d do a good-enough job on the project that people would cheer me on. Eventually, I’d move laterally to another project once the current project felt too “tiresome”. I also wouldn’t do a good job of holding clear boundaries within the project where I did feel some resistance. So I’d continue getting invitations that weren’t entirely energetically aligned with me, I’d then push down this resistance to create calm via meditative practice, and the cycle would repeat. Consequently, each emotional outburst would gradually be greater than the last.

For example, it sounds ludicrous now, but a friend actually needed to convince me to raise my hand to volunteer for Gemini. So I started on the original pre-training sprint, but got rapidly absorbed into drama triangles within that sprint. So I shifted to post-training on Bard, but the overall center of mass of drama moved there instead. I eventually found myself on the instruction following team in NYC. I built a project that went viral within Gemini and recruited a few people to spend cycles on it, but I refused to actually lead them. I was too scared to hold clearer boundaries, and to concretely articulate why I was excited about that project in the first place. So people, including my management chain, projected their own visions and purposes onto the project. This triggered deep internal resistance. I refused to create and delegate concrete tasks, and refused to enroll people into the broader purpose of what we were building. This contributed to the overall drama within Gemini, which generated even more resistance in my body.

LLMs profoundly reshaped the structure of the ML ecosystem. This change led to substantial drama in their wake. The temperature just kept rising within each passing quarter. I wanted to work on LLMs to “change the world” and because they fascinated me, but I also just wanted to be left alone from the drama. Each passing quarter made the idea of being within any frontier lab, not just Gemini, harder and harder.

The decision to stay or leave wasn’t just about work. It had deep implications for an entire life I wasn’t sure I was ready to dismantle. The “stability” of a big company was useful for my relationship with my girlfriend at the time, along with the overall lifestyle I’d grown used to. It was incredibly difficult to separate “do I want this job?” from “do I want this life?” from “do I want this identity?”. I didn’t have conscious awareness of how interconnected these questions had become. Bringing these questions into consciousness would quickly overwhelm me, so I remained stuck.

I experienced two ruptures in 2024 that forced me to face all of this. I got my green card, and my girlfriend and I broke up. That break-up devastated me. I thought I’d eventually marry her. In conjunction with the green card, it rapidly untethered me from all my previous stories about why I wanted to stay at Google, and all my overall stories of how I’d organized my life. Separately, I’d often told myself that I’d quit to do my own thing once I got the green card. Yet here it was, and I was totally stuck. Why was I still at Google?

This stuckness rapidly infected every other part of my life. Meditative practices that once worked started failing me. Thus far I’d managed to stabilize my emotions via meditative practices like Mindfulness of the Breath, and processing them in therapy. Jarringly, I went from being a regular meditator to being totally unable to do any practice of any sort. It didn’t matter if it was Mindfulness of the Breath, Cultivation of Loving-Kindness, Body Scans, etc. I’d feel a deep resistance in my body whenever I tried.

My stuckness got so bad that I took a three-month mental health break from work to process all this. I’m deeply grateful for how supportive my entire management chain (especially my manager) was during this time. I’d often pay lip service to disconnecting from work during vacations and leave. This time I meant it. I put my laptop on the top shelf of my cupboard so it’d be totally out of sight, and out of mind.

I kept asking myself - why do I feel so lost and confused?

I eventually realized that I desperately wanted to feel alive at work, rather than being instrumentally useful for the company. I wanted my energy to come from purpose rather than stress, fear or judgement. Ditto for the professional environment I was participating in. I wanted fewer “meetings” on my calendar and more “good vibes and fun hangs” with my coworkers. To work on something more than the reactive undercurrent of “beat this other lab on this benchmark”. I yearned for a clarity of purpose in creating specific value for specific people with specific problems. At the bottom, I wished for the capacity to stay centered in the face of drama, and to co-create an environment that had structural mechanisms to rapidly overcome it.

I was stuck at Gemini. I couldn’t leave, and I couldn’t stay. Leaving felt disorienting because I didn’t know what I was leaving for. I’d be jumping off a cliff into a fog of confusion and doubt without any navigational equipment whatsoever. The spaciousness of the world was overwhelming. What did I want outside of my identity at Google? Who did I want to be other than an ML engineer? Moving to another frontier lab wasn’t a viable solution. Sure, some things would have been better at OpenAI or Anthropic, and other things would have been worse. Being at another lab wouldn’t change the fundamental dynamic of working inside a large organization whose purposes were increasingly misaligned with whatever was pulling at me, even if I didn’t know what that thing was. Leaving Google also meant leaving a number of delightful managers and coworkers, especially since I loved their good vibes. My manager, director and VP at the time were such good dudes, and I loved shooting the shit with them.

Staying at Google left me with a pit in my stomach. It meant watching my soul erode before my eyes. I saw my continued participation at Google turning me increasingly avoidant, reactive or numb. Of performing excitement I didn’t feel or slowly becoming someone I didn’t recognize.

My three-month leave was one of the best things that ever happened to me. For the first time in a long time, I had the space to gain awareness on where these deep feelings of resistance were coming from. For example, I vividly remember deeply reading Evan Thompson’s textbook Mind in Life, and feeling so much peace, joy and flow. I couldn’t remember the last time I’d had the opportunity to go so deep on any one particular thing. Especially not in the day-to-day firefighting of the LLM era. No wonder I’d been feeling increasingly resistant to being in a frontier lab.

I also gained awareness on why my meditation practice had started to stall. I was introduced to Charlie Awbery via a mutual connection, who taught me their Opening Awareness practice. I was hooked! I’d never done a practice that was simultaneously so calm, somatic and engaged with the world. I felt very little resistance to doing this practice everyday, on the cushion and off the cushion.

A daily Opening Awareness practice started making me aware of all sorts of things that had been sitting just beneath the surface. I realized just how important uninterrupted purpose-driven work was for my day-to-day experience of joy. And how little of that tone I’d participated in over the last few years. Working through and integrating these awarenesses into my life started rapidly releasing the tension I’d been carrying at the base of my belly, and in my sternum. I hadn’t even realized I’d been carrying this tension.

I was jarred at how far my day-to-day somatic experience at Google was from where I wanted to be. My daily experience within Gemini was one of a million Google Chat rooms with red notifications pinging constantly, everyone constantly rushing, fueled with reactivity, urgency and stress. I was so done with using stress rather than purpose as my fuel source.

I couldn’t see until that moment that I’d spent years confusing calm for clarity. Don’t get me wrong, I’m grateful for the calm and stability that my previous meditation practices brought into my life. Especially since I was in crisis. However, calm and stability aren’t the same as being able to see my situation with stark clarity. I’d been parsing the absence of instability as clear understanding.

Why weren’t my previous meditative practices producing clarity?

All meditative practices work by repeatedly and intentionally directing our attention towards something, until it becomes an unconscious habit. The Mindfulness of the Breath involves repeatedly prioritizing a specific object in our awareness (e.g. the breath), over distractions (e.g. other thoughts, internal bodily sensations, external stimuli like sounds, etc.). Some presentations of the practice may involve gently acknowledging these distractions, but they all generally involve bringing the mind back to a specific object. This is akin to, but not exactly like, thought suppression, where one consciously sets aside feelings, thoughts or impulses, especially at high doses of the practice.

Most typical presentations of mindfulness have such a suppressive result at very high doses. This doesn’t make them “bad”! Such gentle suppression can be constructive for an individual in crisis whose emotions have become unregulated. However, too much can habituate this pattern of suppressing thoughts and feelings, and can ultimately result in generalized avoidance. I’d spent a lot of time (many thousands of hours) doing mindfulness practices, and the resulting calm helped tremendously across all areas of my life. But it came at the cost of clarity.

But hadn’t this practice worked for lots of people in history? Why was it having this result for me? Let’s take Goenka-style mindfulness as an example. The tradition that developed this practice was originally embedded within a series of individual monastic vows and communal monastic norms. The vows provided pre-decided answers to organize every domain of human need from where to eat, live, relational norms, etc. Monks didn’t need to navigate career decisions, romantic relationships, financial planning, etc. like the laity. The vows oriented each monk’s life towards the single developmental through-line of purpose called nibbana. This orientation shaped each monk’s perceptions and worked in lockstep with an overall family of practices, including specific suppressive ones.

I didn’t have such monastic vows nor a rich set of communal norms to organize my development. I was inside Google, inside the US, inside of Capitalism. This was fine so long as Google’s through-line of purpose was aligned with mine. However, the broader purpose orienting me started evolving and growing as I gained emotional stability and did my inner work. Eventually, my internal purpose and Google’s purpose increasingly came into deep conflict. I couldn’t consciously perceive this dissonance, partly because the suppression from my mindfulness practice had already grown into emotional avoidance.

Zooming out, everything I was doing was oriented towards staying in place with stability, rather than creating stark clarity. My therapy had become more about managing my emotional symptoms, and for building self-acceptance to create stability. My behavior at Google was more about avoiding discomfort, rather than asking myself honest questions about what I wanted, and then pursuing it. My analytical problem-solving skills stabilized my framing of my overall situation, rather than critically examining the framing itself.

I didn’t need more stability. I needed to grow my capacity to change. I’ve previously discussed the difference between heroism and authenticity. Heroism is characterized by generating the courage to defend oneself from one’s fears, whereas authenticity is characterized by generating the courage to heal and change based on one’s fears, judgements and pain. I’d been using my meditation practice to relate to my experience heroically. I’d sit down, silence the voice of fear/resistance within me, and walk away thinking I’d solved the problem. Opening Awareness helped me relate to my experience authentically, by creating clarity around what my discomfort was actually pointing at.

The “real” question wasn’t about staying/leaving at Google. It was - could I build the capacity to become aware of what I actually wanted, to have the courage to be oriented towards it, and act from awareness rather than reactivity so that I wouldn’t self-sabotage?

I’d returned from my three-month leave as a totally different person. I had so much clarity on what I wanted. Clarity on the gap between the kind of work I wanted to do, and what was actually possible at Google. Clarity on how I was complicit in creating the conditions in my life that I claimed I didn’t want. I loved the people I worked with but vividly remember, “I don’t belong here anymore”. I was able to sit in these questions, and ultimately make a clean decision to leave. I gave notice a few days later.

I wanted to align myself with the purpose of giving people pathways for the proactive cultivation of wisdom, augmented with AI. This didn’t seem possible at Google in the ways that I wanted. So it was time to go.

Opening Awareness was the core practice which created this shift within me. Unlike Mindfulness of the Breath, it involves relaxing into awareness without proactively choosing any object. Awareness finds whatever it wants to find within the field of perception (e.g. thoughts, feelings, impulses), without judgement or control, but also without getting too “involved” in what’s found. One simply rests in awareness of everything moving around in one’s field of perception.

There’s many similar practices in Buddhism, and this specific practice comes from a Vajrayana/Dzogchen tradition. It doesn’t produce the same suppressive dynamics as Mindfulness of the Breath because its mechanism works in the opposite direction. The entire mechanism involves allowing perception of awareness to widen, rather than proactively narrowing it, so there’s nothing to suppress in the same way.

A regular practice of Opening Awareness started pulling me towards other awareness-increasing practices. For example, doing the Morning Pages and associated exercises from The Artist’s Way by Julia Cameron changed my life.

It’s pretty easy to get started with Opening Awareness. The practice itself is profoundly deep and an entire book may only scratch the surface. However, here’s an example to give you a quick taste of how I started doing the practice.

1. I’d find a lively place to sit. Perhaps a park bench, or a noisy part of my home.

2. I’d close my eyes and take some deep breaths. Slowly, in through the nose and out through the nose, making full use of my lung capacity.

3. I’d let everything in my senses flood in, but wouldn’t fixate on the breath. I’d open my hearing awareness to the birds chirping in the trees, the sirens buzzing in the background or people chattering around me. I’d open my bodily awareness to the weight of my body on the bench, any breeze on my face, the warmth on my skin, the pressure of my heartbeat, or the ebb and flow of my chest as I breathed.

4. I’d allow my awareness to move freely in these sensations while remaining “uninvolved”. My awareness might happily flit between the birds chirping in the trees, to my heartbeat, to the warmth on my skin, and back to some people chatting behind me. I wouldn’t worry about getting “distracted”. I’d let it do what it wanted, and wouldn’t try to force it to do anything. Simultaneously, I wouldn’t let it linger too long on any particular perception in my awareness. I’d know I’d gotten “too involved” once I’d lost awareness of everything else.

The crux of Opening Awareness is remaining “uninvolved”. “Uninvolved” doesn’t mean dissociating from experience. Neither does it mean suppressing thoughts, sounds and emotions from experience. It’s an open stance of being aware of being receptive to anything that one finds within their field of awareness. Acknowledging it, but not getting totally swept away by it. Building a habit with this practice made me increasingly aware of the richness and intricacies of my inner monologue. This way of relating to my experience, particularly via meditation, was deeply foreign to me. I needed weekly 1:1 coaching with my teacher Charlie Awbery over a period of months before I even began to “grok” the practice. Charlie was able to diagnose my specific situation, and to offer specific instructions for effectively downloading the practice into my body.

It’s worth noting that Opening Awareness was part of a broader complementary system, rather than a standalone tool. I was already in therapy and used those sessions to integrate any difficult emotions that came up during this practice. I was simultaneously working with a psychiatrist to experiment with ADHD medication for the first time. I’d already been on SSRIs for a while, but finding a good stimulant helped me build a consistent routine for this practice without relying on adrenaline, stress or fear as fuel.

Integrating what came up from this practice released so much chronic tension in my body. I started sleeping better, and had a reliable bedtime for the first time in my life. I started going to the gym more consistently than I’d ever done, and started cooking healthier meals at home.

I gradually found so much more peace, joy, love and flow in my life. After leaving, it was so much easier to find the right purpose-driven projects that gave me money and meaning simultaneously. I found a community that was in far greater alignment with me, and I didn’t feel as much chronic stress on work projects anymore.

I’m so grateful for everything and everyone from Google, but I’m glad that I had the clarity to leave. I feel like more of who I truly am with each passing day since leaving.

If any of this resonated with you, and you’d like to gain clarity on whether you should leave your job, please feel free to reach out at varun@doubleascent.com. Write to Charlie Awbery at coaching@awbery.com, who taught me this practice.

Further Reading

• Charlie Awbery’s Opening Awareness book, and blog post on the practice.

• David Chapman’s book at vividness.live if you’d like to learn more about Dzogchen.

Acknowledgements

• Charlie Awbery for teaching me Opening Awareness, and for helping me understand how the different pieces of my practice life fit together.

• Brian Basham for reconnecting with me during my leave, teaching me his emotional surfing practice, and suggesting I connect with Charlie.

• My therapist for everything she’s done for me. Including giving me the homework of leaving my apartment at least once a day during my leave.

• Professor John Vervaeke for everything he’s taught me over the years.

I kept getting promoted at Google while being unable to turn on my stove.

The promotions made sense to anyone watching from the outside. The stove made no sense to anyone, including me. I’d stand in my kitchen, paralyzed by intrusive thoughts of burning down my apartment. I’d obsessively wash my hands before eating. I’d check and re-check the door lock. I couldn’t pay my bills on time, couldn’t take out the trash consistently, couldn’t keep commitments I’d made to myself.

That last one defined my life. I’d make a considered, values-aligned commitment on a Sunday, and by Thursday I’d have forgotten it existed. Not failed to do it. Forgotten it existed. As if the thought had never happened. I tried to-do lists, Pomodoro, “mindfulness” practices. They’d last days at most before the same pattern swallowed them whole. I’d lived this way for as long as I could remember, and I’d accumulated a proportionate amount of shame along the way. My entire life was organized around anxiety and stress as fuel sources.

I was eventually diagnosed with ADHD and OCD as an adult. This is a story of a practice I used to rapidly bring my symptoms under control, and live a fuller live. If you can relate to the symptoms below, I hope this story encourages you.

ADHD/OCD are just labels. It’s okay if you don’t resonate with them. This essay may still be for you if you resonate with the symptoms described below. Nothing here is medical advice, although if any of it resonates I’d recommend talking to a qualified therapist.

My symptoms had a clear pattern that would start with me living my life in a good mood until some experience triggered an uncomfortable emotion. These triggers were many and varied, and included everything from the anxiety of turning on a stove to feeling “rejected” in a work meeting. I’d be especially vulnerable if I was already feeling lonely or misunderstood. It wouldn’t even have to be an extreme situation, but rather one that crossed some invisible threshold that changed day-over-day and week-over-week. This unpredictability itself eventually became a sort of dread that made self-trust difficult.

Each trigger’s discomfort would lead to an immediate and simultaneous push/pull within my body. I’d get pushed away from fully experiencing the uncomfortable emotion. It’d be as if I’d get set on fire if I actually sat down to experience it. Simultaneously, I’d get pulled towards some self-soothing compulsion. These included binging on refined sugar, social media, alcohol, checking and re-checking things or vomiting my feelings on someone else.

My brain would get “locked” towards the compulsion’s object. My body would get autonomously possessed and move towards it. It was obvious these compulsions were self-destructive, but I felt utterly powerless in their wake. They brought forth unending self-hatred and shame. I especially hated myself for the times these compulsions brought me joy. To this day, I love chocolate chip cookies. Fortunately, they’re no longer a “problem” for me.

Not feeding the compulsion was intolerable. The more I resisted, the more brutal my anxiety and discomfort would become. I have vivid memories from my Sydney apartment, of crouching in the corner in the darkness waiting for the emotional storm to pass. Grim with the knowledge that the following day would be another war with myself.

Indulging each compulsion would usually have some immediate negative consequence that would narrow my agency the following day. An upset stomach from binge eating, a hangover, wasted time, insomnia, etc. I’d then be that much more sensitive to the subsequent trigger.

It’d all eventually add up and I’d “crash”. I’d have lower-than-normal productivity at work, avoid friends, hermit in my apartment, etc. Unfortunately, I wouldn’t get clean rest when I crashed. I’d shield myself from the world but fill that time with things that weren’t genuinely restorative. I’d stay up late binging on TV or playing video games. I’d eat whatever required the least work even if it wasn’t good for me. I was hiding from the triggering stimuli but I wasn’t recovering, I was surviving.

My energy would eventually trickle back not because I’d healed, but because I’d hidden long enough for the acute distress to subside. Unfortunately, this “recovery” was often accompanied by shame. “This isn’t me. I can be better than this.” That shame would turn into self-contempt, fueling a resolution to find a habit or skill to finally fix myself. Predictably, I’d get triggered by something new and once again it’d all go to shit.

I’d bleed self-trust during each round of this cycle, and the world increasingly became hollower and paler. I stopped trusting myself, other people and eventually lost the ability to truly trust anyone or even hope itself. I internalized the idea that I was lazy, stupid, weird and crazy. My nervous system was like a racecar that only had two speeds, stationary and Mach 7. It had no ability to do gentle right turns without careening off the road. It’d get totally derailed by a small pebble on the road.

The only organizing force in my life were my external commitments. I held down a demanding job doing deep learning research at Google by wielding anxiety, stress and adrenaline, which exacerbated my insecurities and anxieties.

It wasn’t all bad. I developed lots of compensatory skills that were professionally useful. Having essentially no executive control over my body forced me to improve my ability to prioritize. I got really good at ingesting lots of information to find the one high-leverage action for the week, even if everything else burned, because I could only do one action a week. This allowed me to consistently make “magical” bets at work. It looked like I was thriving on the outside, but on the inside I was held together by duct tape and chicken wire.

I could see all the cool things I’d do with my life if only I could figure out how to do more than one thing a week. Forget about finding meaning in life or following a deeper purpose. I wanted to convince myself that I was capable of more than survival. To prove to myself and the world that I was actually worth something, and not the piece of shit I feared I actually was. I wanted to publish high-impact research in AI, get jacked, find a loving girlfriend and have meaningful friendships. I’d spent my whole life surviving and I was so tired.

I was torn between two competing narratives. The first was that I wasn’t trying hard enough because I was acting lazy, stupid, weird, unstable or crazy. That I was a mess because I simply didn’t care enough. That I needed to buckle down, essentially the way every adult had scolded me for since I was a child. I was already trying as hard as I could and nothing was working, so this option brought nothing but despair. The second was to give up and accept that not only was my life a mess but it’d always be a mess, because I was inherently broken. That the adults from my childhood were right and I’d always be a fuck-up or broken in the head.

Both narratives simultaneously fueled my downward spiral. Trying harder was the on-phase of the cycle where I believed some specific skill would save me. Giving up was the off-phase of the cycle where I recovered from the crash. Neither option gave me any real agency, or changed things in any meaningful way.

I’d lived this way all my life until I started therapy at the age of 29 with the right therapist. She changed my life, and I’ll be forever grateful. Everything good in my life from that point on, has been due to the space she held for me during my crisis.

Therapy gave me the official ADHD/OCD diagnosis, helped me see the deeper wounds driving this process and gave me more clarity on what was actually happening.

There was an asymmetry between my internal and external commitments. My issues were especially pronounced when keeping commitments I’d made with myself rather than commitments I’d made with others. Social sanctioning was terrifying and extremely organizing for a nervous system hypersensitive to rejection and abandonment. This terror would easily override any other fears blocking me from meeting my external commitments. Internal commitments provided no such overrides so they’d never get respected. This wasn’t purely about childhood abandonment wounds, although those were significant. My nervous system was also likely more sensitive than the average person’s to begin with. That is, it was predisposed to sound the alarm far sooner than the average person’s.

Every failure to overcome the alarm would reinforce the fears that produced the alarm. This caused my nervous system to rapidly lose all proportionate calibration for fear. So my life rapidly conformed to revolve around it. I was in a permanent state of hyper-vigilance making me extra susceptible to triggering the downward spirals. My world eventually narrowed until I was largely driven by the fears of social abandonment if I didn’t meet my external commitments, and the fear of “wasting my life” if I didn’t have external commitments because I fundamentally lacked self-trust and self-worth.

This is why none of my one-off interventions worked. Shallow therapy via Lyra gave me emotional support, but lacked insight and support for changing my observable behavior. Meditation temporarily created calm but didn’t change the underlying fear dynamics. So a skipped day would bring everything back. “Eating the frog” demanded I confront my biggest fear first, which was inappropriate for a nervous system prone to over-triggering.

Moreover, none of these interventions worked with my shame. Shame was the cycle’s fuel, not the byproduct. My “recovery” was actually shame performing as hope, and the energy to “try again” came from self-contempt rather than self-trust. That’s exactly why it was so brittle. The compensatory skills that were rewarded at work further amplified my shame by reinforcing the idea that the unmasked version of me was unacceptable. I couldn’t build self-trust while masking who I was because every success belonged to the performance, not to me. The shame led me to hide, and prevented exactly the kind of honest contact with fear that could have changed things.

At root, my fear response was incredibly miscalibrated. I’d built my entire life around avoiding the alarm rather than building the capacity to question whether the alarm was telling the truth.

I wasn’t lazy, stupid or crazy. I was in survival mode, spending every ounce of energy I had to get through the day with a foghorn of an alarm constantly blaring in the background.

I needed a shift from survival to heroism. Specifically, the psychologically developmental capacity to defend against fear’s disorganizing effects and act with intentionality despite them. The quiet, daily heroism of choosing to face what scared me and acting skillfully in its presence.

I couldn’t build heroism when my entire bandwidth was consumed by survival. Every one-off skill I tried assumed bandwidth that simply didn’t exist. I needed a meta-skill that increased my capacity to defend against fear’s disorganizing effects, while being robust to wild swings in my bandwidth.

Every specific skill I’d tried thus far presented itself as a finite game. Breaking a streak or completing a goal would evaporate its motivational energy. I needed an infinite game with a clear curriculum of sub-goals that would provide motivation for the next thing the moment a sub-goal was completed.

Every avoided task sat atop a fear. I use “fear” broadly here to capture the felt sense of some internal resistance, and the quality of being energetically or unconsciously repelled from certain tasks. I realized that these fears were cross-cutting across all the domains of my life. It was the same nervous system underneath avoiding the difficult email, a difficult gym session or a difficult conversation. The set of all my avoidances was the curriculum I needed for my infinite game.

I needed to cultivate the meta-skill of systematically facing my avoidances, from the easiest to the hardest, to “master” the infinite game of growing braver.

This flipped everything on its head. The dopamine-hungry nervous system that once chased compulsions with ferocity started to chase the positive goal of psychological development with that same ferocity! The goal wasn’t to suppress my superpowers, but rather to redirect them towards some positive purpose.

“Eating the frog” had the right insight of regularly facing avoidance, but got the curriculum wrong for my nervous system. My practice did something similar, but with the right curriculum.

Within nine months I went from being unable to turn on the stove to keeping a clean apartment, reliably engaging with challenging conversations at work, holding responsibility, expressing intimacy and gained the courage to move to NYC.

With that said, let’s take a step back from my story to understand the mechanics of how the practice works, and why it works.

The practice is not complicated to state:

1. I’d find a quiet moment in my day, sit down, put my phone away, take a few deep breaths to calm myself. Then I’d close my eyes and ask myself honestly - “What are all the things I’m avoiding which, if I stopped avoiding, my life would improve?”

2. My mind would answer without restraint when asked calmly and earnestly. I’d usually get hit with a flood of tasks.

3. I’d acknowledge each task that came up, constantly asking “What else?” until nothing else came up. Each task would have some difficulty on a scale of 1-10, and I’d compassionately acknowledge each one.

4. I’d sit and reflect until I found a task at 4/10 difficulty.

   • 4/10 is what my nervous system found to be both safe and scary.

   • Note that I’d adjust what 4/10 meant based on my nervous system at the moment when I asked the question in step 1.

     • Some days 4/10 involved doing the laundry or eating slightly healthier food.

     • Other days 4/10 involved having a very charged crucial conversation at work.

   • This is a relative rating scale, rather than an absolute one. It’s benchmarked against what I found “easiest” and “hardest” on the specific day that I asked this question.

5. I’d make a contract with myself to not fall asleep until I’d finished the 4/10 task. I’d hype myself up and congratulate myself whenever I completed the task.

6. I’d lower the threshold for every 5 contiguous days that I couldn’t sustain the practice. I’d increase the threshold for every 5-day streak. I’d never exceed the threshold above 4/10.

7. If a particular task showed up again and again, or whenever the threshold dropped below 1/10, I’d discuss the specific obstacle with my therapist.

This practice has two functions - recalibration and diagnosis. Each successful daily contact with fear provides experiential data to recalibrate the fear response. This is what compounds over time. Steps 5 and 6 were a diagnostic mirror. They helped me practice adjusting my threshold, and building self-trust on that day’s capacity. They’d inevitably shine a light on shame, resentment, frustration, etc. that kept fueling my downward spirals, but at a level of scope that wasn’t overwhelming.

Discussing these internal resistances with my therapist was often extremely efficient. Especially because the practice’s diagnostic function continuously generated a stream of specific, observable targets. “I keep trying to write this email and I just can’t do it” is enormously more useful to bring to a therapist than “I have trouble with avoidance”. The practice narrowed the search space from “everything that’s wrong with me” to “this specific thing, on this specific day, at this specific threshold is hard to accomplish”. That specificity transformed and rapidly accelerated my therapeutic work.

In that sense, this practice is part of a broader system rather than a standalone tool. I didn’t do it practice in a vacuum. I was in therapy, on medication and was rapidly developing a healthy community around me. It all operated in a gigantic positive feedback loop.

Understanding the principles below will help you customize the practice to your needs.

• The low starting dose bypassed my miscalibrated machinery.

  • Sudden changes in bandwidth didn’t matter if the dose was low enough, and benchmarked to my current conditions.

  • In the limit, if I didn’t have the energy to actually do any of the tasks, I’d just sit there and visualize myself successfully completing the task.

• Consistency over intensity.

  • A smaller dose every day is better than a larger dose three times a week.

  • Also note that the easiest tasks weren’t “easy”. They were actually the “easiest” tasks from the “hardest” tasks in my life, because they were tasks I was systematically avoiding.

• Calibrating to that day’s nervous system.

  • The “difficulty rating” of each task was a moving target. Yesterday’s 3/10 might be today’s 7/10 because I slept badly last night, or was in the tail end of a crash. The practice adapted to me, rather than the other way around and always gave me something valuable to do.

  • Everyday was an opportunity to practice calibrating myself with what was sustainable for my body.

• The gains compound.

  • Doing the practice more often caused the body to do the practice unbidden outside my conscious “practice time”. This would only reinforce the behavior.

  • On an absolute scale the practice ebbed and flowed, but it rapidly compounded because there was always some non-zero dose fed into my system every day. And everything operated in feedback with everything else.

  • Therapy, medication and supportive people around me both fed the practice, and were fed by the practice.

• It’s cross-cutting.

  • The core fears driving avoidance in one domain often share similar roots with another domain. They often share similar underlying stories.

  • The practice is holistic by design because it asked what I’m avoiding across the scope of my entire life. It’d provide an organic curriculum cutting through every dimension of my life.

I wasn’t on medication when I started this practice, but getting onto a daily SSRI substantially accelerated my progress. I had substantial initial aversion to being on medication, but on balance, I’m glad I went on them. The SSRI gave me enough stability to allow a practice like this to substantially widen my field of agency. Starting a stimulant for my ADHD further accelerated the potency of this practice. It was incredibly helpful to work with a good psychiatrist to consider the broader implications of being on medication. This is yet another reminder that this practice wasn’t a standalone tool, but part of a broader system. This practice’s role was to act as an integrator for all the different interventions within my life.

I still experience intrusive thoughts and triggers, and I’m not perfect. However, they don’t govern my life the way they once did.

My fears have now become my friends, and actively help me skillfully navigate challenging decisions. However, I first needed to get out of survival into heroism before I’d have been able to authentically befriend my fears.

Despite the suffering, I’m grateful for what I went through. This journey was a gift in showing me how to develop conscious competence in growing my capacity for courage, and many other psychological traits. Also, I did get everything I wanted in the end. I worked on some exciting AI projects, got fitter, had a loving and meaningful romantic relationship and rebuilt my life with deep self-trust. The version of me from 2021 wouldn’t recognize who I am now.

If any of this resonated with you, please feel to reach varun@doubleascent.com or drop a comment below.

Disclaimer

Nothing in this essay constitutes medical advice. I’m sharing my personal experience with practices that helped me. Please consult qualified medical and mental health professionals for your specific situation.

Acknowledgements

• I’m deeply grateful to my therapist for everything she’s done for me.

• The following lecture series had a profound impact on me and helped me formulate this practice:

  • John Vervaeke’s Awakening from the Meaning Crisis

  • Jordan Peterson’s Maps of Meaning

  • Jordan Peterson’s Personality and its Transformations

• The following books were invaluable:

  • Brain Lock by Jeffrey Schwartz

  • You Mean I’m Not Lazy, Stupid or Crazy?!: The Classic Self-Help Book for Adults with Attention Deficit Disorder by Kate Kelly, Peggy Ramundo and Edward Hallowell

• Brian Whetten and Charlie Awbury for improving the clarity around the ideas discussed here.

• Calvin Nguyen for helpful feedback on an earlier version of this post.

However, those essays left an important question unanswered. If developmental problems are ubiquitous, which ones are worth prioritizing? You’re finite in space and time. It’ll never be possible to free yourself from all systematic errors in your cognition.

So then, what’s the overall process of psychological development when viewed as a whole, rather than one wall at a time?

More specifically, how can I develop the competence to navigate the process of psychological development deliberately, rather than leaving it to the shifting winds of “Fate”?

Why existing pathways don’t work for me

I’ve practiced various forms of Buddhism for years. I find deep wisdom and inspiration in the Bible, the Gita, and the overall Mahabharata. However, I primarily relate to them through a literary and metaphorical lens. I don’t look to them for scientific truths.

I see the value they provide to billions across the world. I also see the through-lines that cut across them. Nevertheless, they’re not existentially viable for me because committing to any of them evokes deep fears in my body. I’ve come to see wisdom with leaving these fears intact, but also the positive purpose of provoking me to find a better relationship to these traditions, that fits me. It’s cold out there, and I need a jacket. I’m not smart enough to tailor one from scratch fast enough before I die, but neither do I want to buy off the rack. I’m left with buying something off the rack, and developing the skill of altering it to fit my body.

Committing to a framework whose propositional truth claims (e.g. afterlife, heaven, rebirth, etc.) strain under the scrutiny of modern science frightens me because I risk intellectual dishonesty I’m unwilling to accept.

Deep devotional commitment to a guru (e.g. guru yoga) frightens me because I risk compromising my autonomy. There’s many instances of guru-student abuse which leave me wary.

Inhabiting a cosmology that doesn’t resonate with the deeper story I actually live inside frightens me because I risk feeling inauthentic, resulting in increased loneliness. Some practices don’t land for me not because the cosmology is bad or wrong, but because I live in a very different world (i.e. Capitalism) than the one in which they were invented centuries ago.

The rational fear of getting burnt by an open flame shouldn’t lead to a general prohibition on cooking food. The proportionate response is to develop conscious competence around wielding fire safely and effectively.

Yet the narrative I live inside (i.e. Capitalism, Science, Modernity, etc.) lack their own native pathways for development. I still yearn for growth and the cultivation of wisdom! I still face developmental walls and competing values.

I’m torn between my fears of committing to any single religious tradition and a genuine yearning for the personal transformation those traditions offer.

A very brief history of religion

Human beings are geared towards growth, even if it happens unconsciously. Humanity has gradually brought developmental patterns into consciousness. Sages and storytellers observed these patterns and codified them into myths and stories best fitted to their own time and place. The Hero’s Journey articulates a through-line running across many of these myths and stories across the ages.

Contemplative and spiritual traditions across civilizations took this much further by systematizing developmental patterns into extraordinarily precise frameworks and institutions. Examples include Buddhaghosa’s Visuddhimagga, various Tibetan Buddhist lam rims, Abhidharmakosa, Patanjali’s Yoga Sutras, etc. from the East, and traditions like Neoplatonism from the West.

These traditions operated through conscious competence, although embedded within specific cosmological frameworks and deeply interwoven with the cultural and political dynamics of their time. Their most sophisticated teachers had a deep, articulated understanding of developmental processes. They could diagnose where practitioners were stuck, and could prescribe specific interventions with relative precision. Most laypeople, and perhaps many of the clergy, lacked deep access to this conscious competence.

Many of these developmental traditions share striking structural similarities despite varying degrees of historical contact. This suggests genuine affordances inherent within human consciousness that different traditions discovered and activated. However, each tradition’s developmental principles are entangled with cosmological claims that may not be necessary for the mechanism of development to function. This entanglement makes it difficult both to compose across traditions and to integrate with the meta-stories of Capitalism and Science.

At the same time, I can’t easily “go backwards” into these traditions by saying that Capitalism is “bad” and that the religions are “good”. Capitalism has become the dominant religion in the world for a positive reason. A primary purpose of any religion is to give its adherents a mechanism for resolving the competing values that constitute each individual’s developmental walls. Capitalism allows individuals to flatten and quantify intangible value into money, which can then be used within markets to resolve competing values through exchange. It’s taken over the world because it enables faster, simpler, and far more decentralized trade-offs between competing values than any previous system.

However, the things I value most (e.g. love, meaning, belonging, growth, etc.) resist clean quantification. Reducing them to monetary terms often strips away the very qualities that make them meaningful. Similarly, Science provides extraordinary material understanding but can’t speak to questions of meaning, purpose and value.

I want to participate in a Way of Being that integrates Capitalism, Science and Development towards what’s Good, True and Beautiful, that’s actually existentially viable for me.

Conscious competence at psychological development

Unconscious incompetence means you don’t know what you don’t know. Conscious incompetence means you recognize what you can’t yet do. I don’t want either of these.

Unconscious competence means the skill has become second nature and no longer requires deliberate attention.

Conscious competence means you can perform the skill with deliberate attention and can articulate what you’re doing and why. The “what you’re doing and why” is critical, and makes it possible to disseminate the skill and to consciously improve it.

The conscious competence of traditional spiritual systems was largely context-specific to its place, time and culture. I can’t go back to the context-specific developmental pathways of traditional spiritual systems, and that ship has sailed. I can, however, cultivate conscious competence at psychological development that is composable with the best truths of these spiritual traditions. A prerequisite is to:

• articulate the developmental affordances for the relevant practices across traditions without buying into their specific cosmology.

• compose practices from different traditions based on what an individual needs at a given stage.

• integrate this developmental work with the realities of modern life rather than requiring a withdrawal from them.

• engage in “scientific” experimentation with an emphasis on “results”.

Every practice has a base, a method, and a result (inspired by this post by Charlie Awbury and David Chapman). The base is the practitioner’s lived experience, their patterns of interaction, their social milieu, and their existing developmental shape. The method is the specific thing one does with their body and mind. This could be a meditative technique, a mantra, a journaling practice, a visualization, an asana, etc. The result is the developmental outcome of the practitioner overcoming some systematic error in their meaning-making.

The translation of language offers a fitting analogy for what I’m pointing at here. Producing a word-for-word substitution across languages isn’t often the goal. Rather, the goal is to capture the original text’s nuance and to express it in the most natural way within the target language. Effective translation requires a subtle understanding of both source and target contexts. In that sense, the composition of developmental practices across traditions is more like skilled translation than transplanting an identical screwdriver from one toolbox to another.

The composition I’m describing here rejects the two extremes of McMindfulness, and the arrogant dismissal of existing traditions.

Some discomfort is likely irreducible within the process of psychological development. However, developing without conscious competence seems inefficient because you waste energy needlessly thrashing against developmental walls, or engaging with walls that aren’t actually a priority.

Participating in a Way of Being that integrates Capitalism, Science and Development towards what’s Good, True and Beautiful, that’s actually existentially viable for me, requires conscious competence at developing psychologically.

Aligning practices with purpose

The base/method/result framework described above captures the mechanism of each individual practice, but traditions are far more than an arbitrary collection of practices. They’re tied together through an embedded purpose, or through-line, which directs the trajectory of a practitioner’s development. For example, the practices within Theravada Buddhism aren’t somehow detached from the ultimate goal of achieving nibbana. Attempting to copy the base/method/result without understanding the underlying purpose risks dissonance between the practice’s original orientation and the practitioner’s actual aims.

I came to meditation through Theravada Buddhism. I gradually understood how vipassana worked in terms of base/method/result, and it made sense to me. However, the deeper I went into the practice, the more I experienced an unconscious resistance. My friction stemmed from a fundamental disagreement between Theravada’s embedded developmental endpoint and my own orientation. I want the capacity to participate in the world with the tone of non-attachment, but I don’t want to materially withdraw into a cave to achieve it. Vajrayana Buddhism (Dzogchen) has become the current best fit for me.

Composable conscious competence requires understanding how the practices function, what they were originally oriented towards, and whether that aligns with the practitioner’s preferred unconscious purposes. Therefore, bringing one’s unconscious purposes to consciousness is a prerequisite for developing conscious competence.

What is “purpose”?

I’ve written about development in terms of cultivating the capacity to navigate greater levels of nebulosity. This can often be reframed as the gradual overcoming of one’s systematic errors. However, humans are limited in space and time, so it’s not possible to overcome all possible systematic errors. That is, it’s not possible to develop “in general”. There’s always some underlying orientation which guides the trajectory of development.

A purpose is a through-line that captures the unity between psychological and physical behaviors at different layers of abstraction. Consider the following hierarchy:

• The specific motor movements within my hand muscles involved with grasping a glass of water.

• These can get “chunked” into groups of motor movements for the broader purpose of picking up the glass.

• These can get chunked into the broader purpose of quenching my thirst.

• This can get chunked into the broader purpose of being “present” on a first date.

• This can get chunked into the broader purpose of growing deeper in a relationship with that woman.

• This can get chunked into a broader purpose of experiencing the fulfillment of romantic love.

• And so on. You get the idea.

Each level of analysis describes a “purpose” that organizes the mind and body underneath it. Purposes are not “out there” in the world, waiting to be discovered. They’re conceptual devices for organizing reality, providing clarity for what would otherwise appear as an overwhelming array of disparate actions and experiences. Different people may see different through-lines of purpose for the same behaviors, depending on their own developmental vantage point. With sufficient integration, it’s possible to track through-lines that cut across systems of people, rather than merely across the the lifetime of a given individual.

Different purposes create different affordances, opportunities, and possibilities for you to meet your various needs. For example, the purpose of “building technology that helps humans grow wiser” would present a very distinct set of developmental walls than “writing an incredible novel that inspires and delights people”. Each through-line of purpose creates both opportunities for flow (e.g. Vervaeke’s description of an insight cascade), and developmental challenges.

The pursuit of purpose is inherently developmental. Purpose-oriented action tends to prioritize foregrounding the specific competing values that need resolution to continue being “on purpose”.

The pursuit of purpose is not “optional”, only conscious or unconscious. Or as Jung put it, “The psychological rule says that when an inner situation is not made conscious, it happens outside, as fate.”

A practice for articulating purpose

Some people can articulate what animates them, and a broader through-line of purpose, with relative ease. Others, like me, need help “sneaking up” on this purpose energy, often for the biggest goals in my life. It’s often easier to start with what states of consciousness I find deeply rewarding, and to use those as entry points for discovering the through-lines that run beneath them.

I’ve found Brian Whetten’s Ideal Scene exercise helpful for “sneaking up” on this purpose energy. It involves articulating a set of present-tense, positively-stated intentions for a given domain of life, and then looking for the through-line that runs beneath all of them. Here are some examples of intentions that have animated me as I understand the broader purpose running through my romantic life:

• I’m peacefully surrendering into the profoundly deep knowing that I am seen, accepted and loved for who I am, without having to perform or prove anything, and my whole body is softening into that trust.

• I’m courageously staying centered and honest through conflict with her, and feeling both of us grow closer and wiser on the other side.

• I’m gratefully relying on her steady, grounded competence and showing up with my own steadiness in return.

Each intention has to be both believable and aspirational, and often has the structure “I’m [emotionally charged adverb] [active verb-ing] [with/in/through] [their quality or shared quality].”

Underlying all of these intentions is the master intention of “This or Something Better for the Highest Good of All Concerned”. This master intention orients purpose towards something larger than pure self-interest without prescribing what that larger thing must be. It sets the stage for what John Vervaeke would call reciprocal opening rather than reciprocal narrowing.

The Ideal Scene is provisional by design. It’s not a permanent life plan but a “good enough” starting point that can be refined through the developmental process itself.

The core loop of consciously undergoing development

The following loop describes the core mechanics for psychological development:

1. Articulate a provisional purpose. This can either be through an Ideal Scene or a Big Goal you already have access to.

2. Create small, safe goals that pull you towards that purpose. Translate the through-line of purpose into concrete, achievable goals that move you in its direction. These goals should be small enough to be actionable and safe enough that failure won’t be catastrophic, but meaningful enough that pursuing them will surface whatever developmental work is needed. These should be “Yes Yes Hell No” goals in Brian Whetten’s language.

3. Notice the developmental walls that emerge. Purpose-oriented goals trigger fear because they tend to go against the grain of existing habits. That is, they surface competing values that now need to be resolved. A core practice at this step is to identify whether a given problem is merely hard, or whether it’s developmental.

4. Resolve the developmental dilemma authentically. A core practice at this step is the one presented in my previous essay. Rather than heroically powering through the wall, you trace back through the fears, judgments, and pain to upgrade the stories underneath each side of the dilemma. This recalibrates your meaning-making system, changes your relationship to the wall and presents a “third choice” to the dilemma.

5. Refine your sense of purpose. Getting to the other side of the wall creates a state of flow, sometimes producing what Vervaeke would call an “insight cascade” where multiple previously stuck areas begin to move simultaneously. You take action from this expanded vantage point, and the new perspective may shift the way you relate to the purpose that oriented you towards the wall in the first place. Some aspects may feel more alive and vivid than before, and others may fall away as no longer relevant. You take this opportunity to sharpen your sense of purpose.

6. Repeat the loop with refined purpose. Each iteration of the loop compounds, bringing ever greater levels of peace, joy, love, and flow.

Purpose is transjective, as John Vervaeke puts it. It’s a real relationship co-created between you and your environment, and it’s neither purely subjective nor purely objective. Each iteration of the loop increases your sensitivity and ability to “find” through-lines of purpose with ever greater scope, such as lines of purpose that run through many people simultaneously.

Each step of this loop triggers fear. Articulating a purpose means committing to a specific direction while accepting that others will receive less attention. Setting goals invites accountability and the possibility of failure. Authentically engaging with a wall means unearthing the fears and pain underneath it. Doing developmental work invites you to change at a deep level. Participating on the other side of the wall is scary because the world looks different now, the old comforts are gone, and you can’t go back.

On the other hand, each step is deeply alive. Articulating a purpose generates substantial “clean” energy and direction that may not have existed before. Goals at your developmental edge produce a distinct quality of excitement that feel enlivening. Facing and accepting a wall often releases tension because it feels “real”, and getting to the other side often produces peace, joy, love, and flow. It invites a sense of spaciousness, gratitude, and renewed purpose that makes the next fear worth facing.

The conscious pursuit of psychological development is scary because it promises to deliver what you truly want. Once you’ve found what you truly want, the thought of losing it is unbearable. Having children often introduces depths of love to parents that they’d never experienced before. On the other hand, they often experience levels of fear they’d previously never experienced.

The more alive and purposeful your life becomes, the more you have to lose, and the more courage is required to continue. The only “solution” is to continue developing further.

Each step of the developmental loop can be augmented by practices from various contemplative traditions, depending on the circumstances of the practitioner. For example, the Ideal Scene can function like a mantra practice or a provocative vow practice; various energetic practices from Vajrayana Buddhism can be used to find and release fears, judgments, and pain. This specific articulation of the loop makes it easy to import and connect with practices across multiple traditions.

Towards developmental engineering?

I believe that the process of psychological development can be made more conscious, in a way that composes practices from various spiritual traditions. I also believe there should be people who specialize in exactly this kind of work. I’m tentatively calling them Developmental Engineers, although I’m still sharpening my thinking around this.

Developmental Engineers would help individuals identify which purposes are most alive for them, and would design tailored practices, tools, and containers to accelerate development towards those purposes as consciously, safely, and effectively as possible. They’d do all this with the composable conscious competence described above, while remaining attentive to the client’s broader social context. Once developmental processes have been brought into consciousness and somewhat systematized, it becomes possible to bring the tools of AI to bear on them. AI is poised to profoundly reshape knowledge work, and developmental engineering is no exception.

A developmental engineer is not a therapist. Therapists hold space for healing, trauma processing, emotional regulation, and many other functions broader than explicit developmental work, and sometimes prerequisites to it.

A developmental engineer is also not a generic executive coach. Executive coaching is broad, encompassing public speaking skills, structural management of organizations, and leadership communication. Many coaches provide valuable support, but their focus isn’t always on the systematic cultivation of psychological development.

A developmental engineer is not a spiritual teacher. Spiritual teachers often operate within a specific tradition and may prescribe purposes based on that tradition’s cosmology. In contrast, a developmental engineer won’t prescribe what purpose a client ought to have.

A developmental engineer may work in tandem with a client’s therapist, coach, and spiritual teacher, but their distinctive function is to afford the cultivation of conscious psychological development, oriented by the purposes most alive for the individual, using whatever practices and tools best fit that person’s situation.

I don’t know yet exactly what the day-to-day of a developmental engineer would look like, but it’s a question I’m deeply interested in.

Please drop a comment below or email me at varun@doubleascent.com if you made it this far and have any thoughts, questions, or feedback.

Acknowledgements

Brian Whetten, Charlie Awbury, John Vervaeke and David Chapman for everything they’ve taught me.

Readers already familiar with Professor John Vervaeke’s work had questions around what makes a developmental dilemma, well, developmental?

As Vervaeke points out, every problem has a corresponding frame. Like him, I use this word to point to something broader than one’s propositional beliefs. It refers to the person’s total pre-reflective/unconscious orientation to a given situation. Within Vervaeke’s nomenclature, I’m referring to the full set of propositional, procedural, perspectival and participatory knowledge which shapes what you find relevant in relation to a situation. More concretely, the things that go into your framing of a situation involve your body’s assessment of safety or threat, the emotional dispositions that color perception before conscious thought begins, and the stories and beliefs that organize raw sensory information into meaning. These are all deeply entangled during the construction of your framing of a problem.

What distinguishes a developmental problem from a merely hard one?

A problem’s overall frame defines what counts as a solution, and once the solution is found it implies how it can be verified with certainty.

Hard problems are solvable in the sense that the frame is stable. The work involves finding the right answer within the frame. The answer can be verified with certainty.

Developmental problems are unsolvable in the sense that the frame itself is unstable and uncertain, and this instability precludes the problem having a “final solution” because such a solution presupposes that the frame itself is “final”. Developmental work isn’t merely about finding the right answer within the right frame. Rather, it involves finding a better frame despite the fact that it’s impossible to find the best frame. In contrast to hard solvable problems, increased clarity and confidence is the best you can hope for with developmental problems.

These two examples cleanly show this distinction.

Calculating whether I have enough money saved up to pay rent for the next twelve months at my current lifestyle is a hard solvable problem. It has a very clear frame with all the relevant variables and constraints, and it has clean verifiable solutions.

In contrast, “Should I leave my stable job at DeepMind for a risky sabbatical?” was a developmental problem. Answering this question involved engaging with a number of solvable problems (e.g. calculating how much money I actually had, projecting how much my current lifestyle would cost over the next year, etc). However, the core difficulty in answering this question arose from an overall framing of the question itself. This framing was built on past pain of taking risks and failing, my overall stories around safety and risk, and an underlying identity that got attached to stability and prestige. This identity got challenged whenever I considered leaving my job because there was an emotional charge of fear, dread, etc. All of this left me stuck. This stuckness was disproportionate to the “objective” reality of pushing buttons on a keyboard to send a resignation email, and subsequently pushing buttons on Google’s HR portal.

The previous essay described a causal sequence of pain generating stories, stories generating judgements, judgements generating fears and fears shaping habits. You then experience a “wall” when these habits pull you in competing directions. This sequence describes the layers of the overall frame that ultimately create the experience of the “wall”. The overall practice within that essay provides an opportunity to peel back the frame’s layers, and upgrade its underlying stories, resulting in a more spacious frame.

Is the distinction between solvable hard problems and developmental problems actually that clear cut?

Unfortunately not. Most solvable hard problems reveal developmental roots when examined more closely. All problems can be framed at multiple levels of depth, with the shallowest depths presenting solvable problems and the deepest depths presenting unsolvable developmental problems.

Here are two examples of problems that aren’t obviously developmental and might even be confused as hard solvable problems. They’re actually developmental if we engage with them closely enough.

Arun’s the CEO of a company and needs to set the company’s strategy for next year. It involves gathering market data, analyzing competitors, assessing the company’s resources, looking at the P&L, etc. So it superficially seems like a hard solvable problem. However, peeling back the frame reveals developmental roots. Arun’s pattern of risk-aversion shapes which strategies feel “realistic” or even “thinkable”. Consequently, the founding team’s unspoken dynamics in relation to Arun shape which directions feel “safe”. These then determine which options actually get serious consideration within their strategy meetings. The final strategy is framed by the developmental shapes of the people in the room. Although it’s primarily shaped by Arun since he has the most power in these meetings. Arriving at a genuinely “better” strategy might require Arun and the executive team to first shift and upgrade their current framing of the situation.

Similarly, Arun has an engineering team that keeps failing to ship a feature. They frame it as a technical problem (i.e. architecture, dependencies, timelines, etc). However, the real issue is that the Tech Lead and PM have fundamentally different visions and neither sees the other’s perspective. The TL previously experienced the pain of production failures due to messy code, and he’s determined to never let that happen again. These experiences frame what a “good product” means to him. The PM previously experienced the pain of losing users to a faster, scrappier startup where slow iteration killed that entire org within the megacorp. Similarly, these experiences frame what “good product” means to him. The PM/TL are genuinely experiencing the competing values of “stable architecture” and “move fast and break things”. Each of their framings strongly filters what feels “realistic”, “thinkable”, “viable” or “safe” to them. Resolving this tension requires resolving the underlying developmental problem, although many hard solvable problems will get solved along the way. This is why buying a fancy new project management tool, writing clearer specs, etc. often don’t work to resolve such tensions. Such tools are solutions to hard solvable problems rather than developmental problems.

The following example makes the nested relationship between hard solvable problems and developmental problems even clearer:

• Layer 1 (hard solvable - technical problem): The codebase has too much tech debt, making changes slow and error-prone.

• Layer 2 (hard solvable - process problem): The codebase is a mess because requirements keep changing mid-sprint, because customer requests get injected directly into the sprint outside formal planning. The team has to scramble and cut corners to keep up.

• Layer 3 (developmental - behavioral problem): Customer requests get injected mid-sprint because Arun personally accepts every customer feature request. He routes them to the VP as a P0 priority. The team isn’t able to push back, and it looks like a behavioral problem between Arun’s and the VP’s dynamic.

• Layer 4 (developmental - identity problem): Arun keeps impulsively saying “yes” because his identity is built around being a leader who “always delivers”. Saying “no” threatens his identity and feels extremely unsafe.

• Layer 5 (developmental - biographical problem): Arun can’t say “no” because he internalized in his childhood that he could only earn love through performance, and transactional value. He has an unconscious story about what he believes makes him worthy of love and belonging, and that makes it hard to say “no”.

Layers 1 and 2 can be treated as hard solvable problems, and solved via technical solutions. However, such solutions only provide temporary relief since the “root cause” is developmental at deeper layers of Arun’s cognition. Fixing developmental problems at their earlier layers only migrates the symptoms elsewhere in the company. Superficially, they often look like totally disconnected problems.

As an aside, Brian Whetten calls such problems “beachballs” where developmental problems are confused for solvable problems. They’re beachballs because the moment you believe you’ve “solved” the problem by pushing it down into the water, it pops up with momentum the moment you let go. The harder you push down, the harder it pops back up. Leadership problems within a company are often beachballs.

All problems are embedded in human contexts. Human contexts are shaped by psychological development, so all problems become developmental if you dig into them deeply enough.

The practical takeaway isn’t that “everything is developmental”. Rather, people systematically underestimate how often the developmental layer is actually what’s most relevant to produce a satisfying outcome.

Whether a problem is developmental depends on the person-problem relationship, neither on the problem nor on the person alone. Different people have different life circumstances and therefore frame the same context differently. Being “unemployed” or taking a “sabbatical” likely means something experientially different to a minimum wage McDonald’s worker, a wealthy tech bro and a billionaire.

The useful distinction isn’t between problems that are developmental and problems that aren’t, but between problems where the developmental layer is relevant right now and problems where it isn’t. A developmental problem’s relevance depends on a person’s current developmental edge and the broader context they’re in. This in itself is a deep question and outside the scope of this essay. I plan on writing more essays on this soon.

If the frames of developmental problems are “the problem”, why doesn’t more information help?

Consider the following questions:

• What are the pros and cons of leaving my job?

• Do I have enough money to weather a market downturn?

• What does my mentor think about me leaving my job?

Such problem-level analysis operates entirely within the frame. Every answer to every question is shaped by the very frame that may need to change. Given that the frame itself may be the obstacle, you often feel more stuck the more you analyze the problem’s content within the frame. Making pro/con lists, gathering more data, etc. are useful tools for hard solvable problems, but may not bring you any closer to shifting the frame itself if they don’t consider the frame’s underlying layers.

Instead, consider the following questions:

• What fears and stories are shaping how I see the decision to quit my job?

• What assumptions about risk am I treating as unquestionable?

• Why does a given option feel unthinkable?

• What judgements am I making about myself or other people whenever I contemplate changing the frame?

Such frame-level analysis makes the frame’s own structure the object of examination. It’s fundamentally different from problem-level analysis even though both involve careful, structured thinking. Your unconscious framing of a situation includes the body’s state (e.g. fears, judgements, pain, etc), and leaving them under the surface impairs the analytical tools that might be available.

This frame-level analysis is exactly what the practice from my last post does. It examines the frame’s own structure by unpeeling these fears, judgements and pain rather than wasting time gathering more information within the frame.

Both frame-level analysis and problem-level analysis are typically required to make progress on a developmental problem. The frame-level analysis provides better questions, and a better framing, within which problem-level analysis can flesh out more information. This then motivates some proportionate forward action, whose resulting outcomes set the stage for the next developmental dilemma and frame-level analysis.

Problem-level analysis often yields clean and “final” solutions, but frame-level analysis defies such finality.

If there’s no “solution” for developmental problems in the traditional sense, what does progress look like?

Progress on developmental problems often yields clarity on the framing, or actions that could improve the framing, rather than yielding “the answers” themselves. More specifically, developmental practices create more spaciousness to ask better questions to oneself, other people or search engines/AIs.

The archetypical challenge of psychological development is that the new frame can’t be fully specified in advance via problem-level analysis. Broader implications of the new frame are discovered “on the fly” via frame-level analysis or retrospectively once you’ve taken actions and new information has started to come in. This is why making developmental progress can feel both scary and exciting. You don’t know what will happen next until you do it.

This reveals another way to distinguish between hard solvable problems and developmental problems. Having the right question is often enough to produce sufficient clarity and forward movement to solve hard problems. The shape of the answer is knowable even if finding it is computationally difficult, because the frame is stable enough once the question is well-formulated. In contrast, even a well-formulated question doesn’t provide sufficient clarity to “solve” a developmental problem because the frame that would define what a “good answer” looks like is itself constantly evolving.

We can see this contrast between actions taken at Level 1 and Level 5 within the example above. The situation within Level 1 lends itself to very stable framings with clear “right answers” once the right question is found. The situation in Level 5 is far more nebulous, and might require substantial iteration until Arun makes progress on his childhood trauma.

Conclusion

Treating a developmental problem as a solvable problem wastes energy and reinforces stuckness.

The first move when encountering a “hard” problem is to ask yourself whether you’re engaging with a hard solvable problem, or a developmental problem.

Acknowledgements

Thank you to Brian Whetten, Charlie Awbury and Professor John Vervaeke for teaching me everything discussed in this post.

You’ve tried making this decision by gathering more knowledge and information. You’ve gathered facts, made lists, analyzed trade-offs, and consulted trusted advisors, but felt no closer to clarity.

You find yourself endlessly ruminating about it in the middle of the day and perhaps the dead of night. There’s something heavier underneath all that rumination. Perhaps it’s grief for the path not taken, or shame that you can’t seem to make decisions that others seem to manage with ease, or perhaps you have a nebulous fear that tightens your chest whenever you try to commit.

Eventually, adrenaline overrides this stuckness. You push down your feelings and pull the trigger on one of the options within the decision. Unfortunately, the decision doesn’t stick, and the stuckness somehow comes back even harder.

Exhausted, you avoid making the decision altogether by coasting or letting circumstances decide for you. This fails to contain the stuckness, and its stressful energy starts bleeding into your work, your relationships, and ultimately your health. You become increasingly irritable, less creative and less present. Every decision that would have felt trivial suddenly feels momentous and produces far more anxiety.

However, this coasting somewhat recharges you in an incredibly stressful way. So once again, your adrenaline attempts to push down your fears to override the stuckness, and once again, it doesn’t really work. Going through these cycles again and again leaves you increasingly hollowed out. Each adrenaline-fueled override leaves you depleted, and each recovery takes longer than the last. The decision sits there unresolved while the rest of your life starts narrowing and shrinking.

You don’t want your inability to make this decision define you. You want to trust yourself to navigate complexity, not because you have all the answers, but because you’d like a more reliable relationship with the process of finding clarity. You’d like hard decisions to arouse curiosity rather than dread, and for each challenge to deepen your sense of meaning rather than erode it. You’d like to live a courageous life with meaning, purpose, impact and flow.

Living a life of such flow isn’t fantasy, and it’s reasonable to want it. However, reaching this life requires a deeper understanding of what’s actually keeping you stuck.

This stuckness is what we’ll call a developmental dilemma. These dilemmas are specifically developmental rather than merely difficult because the stuckness doesn’t originate purely from the external situation. It’s a consequence of limitations in your current way of making meaning, which is shaped by your psychological development. A merely difficult dilemma is one where the situation is genuinely difficult, but you have the internal capacity to navigate its complexity. A developmental dilemma is different. It feels irreconcilable because your current relationship with yourself and the world can’t hold both options with the dilemma simultaneously. Resolving the dilemma requires more than just gathering more data or summoning more willpower. It requires growing your capacity to see more clearly.

Here are some examples of such developmental dilemmas:

• Should I leave the stable job where I’ve built ten years of trust and security for the sabbatical that could transform my career or leave me with nothing?

• Should I stay with someone who loves me steadily but doesn’t ignite me, or end this good-enough relationship and risk being alone?

• Should I stay in the rural town where my extended family needs me and my roots run deep, or should I move across the country to a large cosmopolitan city where I finally feel like myself?

Developmental dilemmas like the ones above reduce to a single archetypal question at the lowest resolution - do I change my way of being, or do I stay the same?

A voice in you clearly sees that your current patterns aren’t working anymore. This recognition is an invitation to change, but it doesn’t feel that clean from the inside. Another voice within you isn’t sure whether the real problem is your approach or the genuine difficulty of the situation. You can’t dismiss either voice because both feel legitimate. Change carries inherent uncertainty and therefore fear, but staying the same feels scary if the environment ends up changing out from under you. A developmental dilemma traps you between the fear of changing, and the perceived cost of not changing at all.

What causes developmental dilemmas?

To understand what’s really going on, let’s look at the underlying machinery.

We’re social primates hardwired to meet various needs. Meeting our needs creates value for us.

However, we may not always have a firm grasp on what our needs are, let alone how to meet them. Nevertheless, we take actions in the world in an attempt to satisfy them. We’re not omniscient, so there’s inevitably some error between where we thought we’d end up, and where we actually end up as we attempt to satisfy our needs. The experience of this error and the gap of the unmet need are sensed as an emotional reaction within our bodies, commonly labelled as pain.

Pain is well...painful! We don’t want more of it!

We make sense of and cope with this pain by creating a story around it. It explains what was responsible for the pain, what it means about us and the world, and implies what we should do differently to avoid it in the future.

This process can happen very quickly. We may not have the resources to carefully, rigorously, and consciously calibrate the story we’ve created in the moment to avoid oversimplifications and distortions. Maybe we’re literally running from a tiger or we’re simply overwhelmed by the complexities of the need we’re attempting to meet. So we cope by repressing our pain and its corresponding story within our unconscious until we have the resources to consciously engage with it.

Unfortunately, repression doesn’t prevent our pain and its story from playing an active role in our lives. On the contrary, they give us unconscious justifications to create judgements within our experience. A judgement is our mind’s attempt to operationalize the underlying story and its unhealed pain. It’s what emerges when the story and our unhealed pain distort a clear-eyed evaluation of reality so as to generate proactively protective emotion. These emotions attempt to protect us from experiencing the same pain again and are often variations of fear.

Fear and judgements aren’t necessarily “bad”, and whether they serve us depends on the context.

Suppose we find ourselves in front of a snarling and hungry tiger sprinting towards us with fangs bared. A snap judgement that arouses fear, which in turn motivates our body to run away, would be deeply adaptive! It’d actually be maladaptive to sit there and contemplate whether the tiger’s actually hungry and wants to eat us, whether our judgement is clouded, etc. That’s how we become tiger food!

On the other hand, instead of a tiger, suppose we have a loving and compassionate girlfriend that genuinely cares about us. Suppose we have ADHD and are really bad at staying on top of housework. We’ve shambled home utterly drained after a long day at work. The moment we step in the door, she tells us that she’s really upset because we haven’t cleaned the apartment the way we agreed we would. It wouldn’t be adaptive to impulsively react to the fear from a snap judgement, even if our nervous system thinks she’s a tiger. The snap judgement doesn’t resolve situations like this, and instead adds more unhelpful energy into the system.

Habits get formed by repeatedly acting out behaviors produced by specific fears and judgements. Habits are our brain’s way of creating shortcuts to get our needs met when we don’t have the mental bandwidth for conscious reflection and consideration. They’re self-reinforcing in that the more we engage in a habit, the stronger it becomes. They’re self-organizing in that the stronger they become, the more “creative” they get at attempting to get their underlying needs met, becoming harder to shake off.

Pain creates stories. Stories create judgements. Judgements create fears. Fears create habits. Competing habits create stuckness.

Consider a fictitious man named Varun who was deciding whether to leave his stable job. He took a risk early in his career that he thought would work, but didn’t pay off, and spent months living with the consequences. That’s his pain. He then unconsciously internalized the story that “taking risks leads to ruin, and the only safe path is the proven one.” It generated judgements that surfaced whenever the idea of quitting his job or taking a sabbatical came up. “I’m upset because I can’t just be satisfied with what I have.” or “I should be grateful for my stability instead of chasing something uncertain”. These judgements produced fear and tightened his chest whenever he imagined himself handing in his resignation. That fear drove one of the habits that kept him stuck. Every time he thought of the sabbatical, he’d whip out his retirement calculator, list all the ways it could go wrong, or ask one more person for advice. None of these actions resolved his dilemma and only served to yank him back to the fork - should I stay or should I leave? That is, should I stay within my existing habits or should I attempt to change?

Our habits and the fears that drive them are the final consequence of our desire to meet our needs based on what we value. Life is complex and we often value multiple things simultaneously. We get stuck in a developmental dilemma when we find ourselves seeking to maximize seemingly competing values of equal importance.

Suppose we’re at a fork and must choose between values A and B. The moment we start walking towards A, the habits fueled by the fears and judgements from B yank us back to the fork. The same thing happens if we start walking towards B. Because habits are self-reinforcing and self-organizing, the more we struggle, the more “creative” they become at pulling us back towards the fork. This causes us to experience our developmental dilemma as if it’s a concrete wall. The harder we push, the more it resists and the more bruised we become from the impact.

Thrashing against the wall merely increases our stress, which in turn worsens our decision-making. This reduces our effectiveness in the world and makes it even harder to meet the needs underneath what we value in A and B. This in turn makes the fears and judgements from both A and B yell out even louder in our minds, which in turn further increases our stress. It’s a vicious cycle that grows and grows until we crash into a developmental crisis where we’ve hit rock bottom.

It’s sometimes possible to power through a developmental dilemma in a way that optimizes both values simultaneously, by white-knuckling the fears on each side. This approach doesn’t actually heal the fears, judgements, and pain underneath the dilemma. So the underlying patterns that generated the dilemma remain intact. Therefore, overcoming the wall in this way immediately presents another wall of the same taste as the previous one, but with far greater severity. We sometimes manage to overpower that one too, so another one gets produced, and so on. Each one requires more energy than the last, because the thrashing required to overpower a wall reinforces the corresponding habits. Eventually, we’re presented with a wall we simply don’t have the resources to override, and our bodies crumple with exhaustion. This is the experience of burnout.

A typical response towards the dilemma in the face of burnout is complete inaction. This is the machinery that produces stagnation and resignation. It eventually manifests as us being “checked out” or “coasting”.

This is also why treating a developmental dilemma as if it’s purely an information-gap problem doesn’t work, because the wall isn’t purely the result of ignorance about the external world. It’s built from miscalibrated unconscious stories about our unhealed pain, the judgements and fears they generate, and the habits that keep pulling us back to the fork. Clarity is achieved not by collecting more external information, but by peeling back the layers of fear, judgements, pain and finally upgrading the miscalibrated story underneath. More information can’t resolve what’s fundamentally a lack of internal integration.

Heroism versus Authenticity

Heroism is this approach of mustering the courage to push through the wall by defending ourselves from our fears. It’s uncomfortable but can generate incredible results because decisions get made and we move forward. Our culture valorizes such heroism and it’s not “bad”. It’s better than the alternative of being a slave to one’s fears. Developing the capacity for heroism is a natural and appropriate part of psychological development.

However, heroism has inherent limits because it doesn’t go deep enough. Pushing through the wall via sheer force of will overrides the relevant habits, but the story, pain, judgements, and fear that generate the habits remain unhealed. Heroism produces short-term results but guarantees an overwhelming escalation in the future.

The alternative to heroism is authenticity.

Relating to your dilemma authentically involves tracing back through the fears, judgements, stories, and pain, and involves a recalibration at the source. We befriend our fears rather than defending against them. We forgive our judgements rather than reacting to them. We heal and grieve our pain instead of ignoring it. Most importantly, we consciously recalibrate and upgrade the story built to make sense of the pain, so that it can see and explain more of reality than when it was first created.

Shifting these deep stories is a profound experience, because we now gain access to a far more expansive view of our dilemma. This improved clarity often produces alternative choices that seem to simultaneously optimize the competing values. Sometimes it even changes the framing of the dilemma itself, by changing what we find relevant in our circumstances. Moreover, it becomes easier to resolve the dilemma consciously if the two competing values are truly mutually exclusive.

Shifting from heroism to authenticity changes our relationship to discomfort so that it informs our choices rather than hijacking them. Overcoming the wall feels less like pushing through it, and more like getting pulled through it.

The practice of cultivating authenticity

The following process is one possible practice for cultivating authenticity. It’s a skill, so it may seem awkward at first but becomes more natural with repetition. It involves running the four steps below for each option in the dilemma, and then integrating them together:

1. Making friends with your fears.

2. Forgiving your self-judgements.

3. Grieving your pain.

4. Upgrading your stories.

Our entry point is fear because that’s what we experience most immediately when we sit with a dilemma. Jung observed that “when an inner situation is not made conscious, it happens outside, as fate”. Rather than heroically powering through our fears, the first move is to slow down and relate to them as an early warning system. We ask our fears questions to better understand what they’re warning us about in the best way that they know how. This shifts our relationship with fear from something to overpower into something to listen to and befriend.

Befriending our fears allows us to see the judgements underneath more clearly. It’s worth distinguishing a judgement from an evaluation. A judgement is what happens when a clear-eyed evaluation gets distorted by pain we haven’t yet healed. Our judgements towards others are ultimately a reflection of self-judgements we haven’t yet owned, because these distortions originate in our own pain. So the next step is to practice compassion to self-forgive each judgement that we’re bringing to the dilemma. Forgiveness doesn’t mean that the judgement is totally devoid of truth. Rather, forgiveness allows us to separate out the evaluation within the judgement from the pain distorting it.

This then creates the space for us to consciously sit with and grieve our pain. We can use our earliest memories of the pain to get a better sense of what aspects of our mind need the most love and healing. Bringing tenderness, care, and compassion to this pain creates the conditions for genuine grieving. Processing this grief loosens the grip the underlying stories have on us, so that they can be upgraded.

Our stories built around our pain were our best attempt at making sense of an overwhelming experience. They were often constructed under duress and haven’t been updated since. Upgrading our stories involves working through a number of journal prompts that progressively widen the aperture of our stories starting with venting/blame, to asking how this entire experience might have been something we uniquely needed. This progression matters because each level of questioning can only land once the previous one has been genuinely processed. For example, trying to skip to “how was this experience for me” before the blame has been heard will feel hollow and forced.

Unpacking each option in the dilemma

The following contains journaling prompts for you to help unpack each option in the dilemma. Each step contains a different theme of journaling prompts which will offer more insight as you progress through the steps.

1. Describe the situation. Map out the choices within the dilemma with some explanation of how you got here. Probe the situation to find all the negative emotions associated with the dilemma. Find all the statements of the form “I’m upset because...” that feel real to you.

2. Explore these feelings.

• What are you feeling about the situation?

• What emotions are coming up for you?

3. Connect to the body. Where in your body can you feel these feelings?

4. Make friends with the fears underneath these feelings. Ask those fears the following questions and engage in a dialog with them.

• What are you scared and/or upset about?

• What are you trying to warn me about? What are you trying to protect me from?

• What are you afraid it would mean if that happened? What is the worst case scenario that you’re worried about?

• What is your positive purpose? How are you serving me the best way you know how?

• In addition to safety, what other positive goals or outcomes are you trying to help me create?

• Give appreciation and acceptance to your voice of fear. Remain connected with the fear via that part of your body, and tell it that “I appreciate your positive purpose, and I’m grateful for how hard you’ve been working to help me meet these goals, the best way you know how.”

• Would you be willing to work together with me, as friends, to move forward with these goals in ways that might feel scary at times, and would also be safe?

• What would you want or need from me, in order to better do so?

• Is there anything else you want to share with me?

• Conclude by genuinely thanking your voice of fear, and praise it for its efforts.

5. Explore any judgements underneath these fears that you’ve placed on the situation, the other person, or yourself. Remember that judgements you hold against other people are often mirrored by judgements you hold against yourself that you haven’t yet owned.

• “I’m upset because...”

• “They should...”

• “I should...”

6. Practice compassionate self-forgiveness. Remember a time when you felt particularly connected and loving. Bring that feeling into your body. Take this feeling and practice compassionate self-forgiveness for any of the judgements you’ve innocently placed on yourself and others. Keep going until you either feel “done” or “lighter”.

• I forgive myself for judging myself as $BLANK.

• I forgive myself for judging $BLANK as $BLANK.

• I forgive myself for judging myself for $BLANK.

• I forgive myself for judging $BLANK for $BLANK.

• I forgive myself for buying into the belief that $BLANK.

7. Find the earliest painful memories. Try to find the earliest painful memories for each judgement. Ask yourself what pain these judgements were attempting to protect you from, in the best way that they knew how. Write down these memories as clearly as you can.

8. Hold compassion for the pain. Practice holding compassion and care for the pain in these memories. Create space to grieve for this pain if you need it.

9. Identify the stories. What are the stories you created to make sense of this pain? Write down all the ones that come up.

10. Upgrade each story through progressively higher levels of thinking. Answer the following questions for each story:

• Why me?

  • This is yet another opportunity to grieve the pain from the story. Give yourself permission to vent whatever you need to vent that you haven’t already vented above.

  • This is also your opportunity to vent about the difficult decision you’ve found yourself in (e.g., if you find it unfair).

• Who is to blame?

  • There will be parts of you that want to ascribe blame. It’s okay that they’re doing this, and they’re trying to help you the best way that they know how. This is your opportunity to give them a voice, even if you’re not going to immediately react to that voice.

  • Figure out who (including you?) was responsible for this pain.

  • Figure out who is responsible within the current context for why you’re facing this difficult decision, and how it seems related to your pain.

• How do I fix it?

  • What would it take for you to internally and externally fix your pain? It’s okay if you don’t have the answers. If you don’t know how, who could you ask for help?

  • What would it take for you to internally and externally fix your pain within the current situation? Again, it’s okay if you don’t have the answers. If so, who could you ask for help?

  • What would it take for you to feel/act with more integrity both within this situation and in relation to your pain?

• How can I learn and heal?

  • Your pain is an experience to learn and grow. What’s the positive lesson behind your pain?

  • What sorts of constructive behaviors can you turn into habits to “bake in” what you’ve learned from this pain?

  • What sorts of constructive behaviors can you engage in to heal from this pain?

  • Again, if you don’t know the answers to these questions, who could you ask for help?

• How was this for me?

  • Beyond learning and healing from the experience, how were both your original pain and this specific dilemma a unique gift for you to practice receiving from the universe?

  • What was the positive purpose of this entire experience, uniquely and idiosyncratically for you?

Integrating the upgraded stories

Going through the steps above for each option in the dilemma will often change your relationship with the dilemma itself. The journal prompts below attempt to integrate all of these insights into concrete decisions you can now make.

11. Explore alternative choices.

• Having upgraded the stories from both arms of the dilemma, were there any core confusions keeping you stuck?

• Are there any possibilities that you can now see that you previously couldn’t, even if they involve asking for help from someone specific?

• Are there any possibilities that would let you grow from this situation?

• What new constructive behaviors would be necessary for you to work towards these new possibilities?

12. Commit to constructive habits.

• Given the constructive behaviors from the last step, how can you establish them as habits? What kind of support do you need to turn them into habits?

This entire process is much easier with the help of a therapist or executive coach. But you can still make meaningful headway by yourself.

Conclusion

Hopefully, your original dilemma will look and feel different after doing this work, not because the external circumstances have changed, but because the fears that were distorting your perception have been recalibrated. In some cases, the dilemma itself may shift such that you view the entire context very differently.

Note that this process isn’t something to be done only once. This process is most valuable when established as a habit that you practice for the difficult decisions that you encounter on your path. Each recalibration doesn’t just resolve this specific dilemma, but rather builds your general capacity to navigate the next one authentically. This practice compounds over time, and what once required step-by-step effort starts to become a more natural way of relating to stuckness.

This practice is a doorway to a life where difficult decisions bring curiosity rather than dread, where each challenge deepens your sense of meaning and clarity rather than eroding it.

Wisdom is the capacity to see more clearly rather than knowing more. Each developmental dilemma is an invitation and opportunity to see more clearly than you could before. In some sense, this practice results in the cultivation of wisdom as practiced through the doorway of difficult decisions.

Acknowledgements

Thank you to Brian Whetten for teaching me everything I’ve articulated in this essay. I cannot emphasize how incredible he is as a coach. All the valuable insights from this essay are his. All mistakes, lack of clarity, etc. are mine.

If you liked this essay, I’d highly recommend reading his book Yes Yes Hell No! The Little Book for Making Big Decisions. There’s an incredible amount of value packed in that one little book.

This essay is especially for you if your insights often unnerve people. If they often call you pessimistic, overly optimistic, arrogant or just ignore you. If whenever you slow down to explain yourself, you lose them because you often need to explain ten other things before you can land a relatively simple insight.

This illegibility eventually infects almost every relationship including the one with your manager. He often can’t understand what you’re saying, so each meeting seems to make him more anxious, which in turn makes you more anxious. His response to not understanding you results in him trying to micromanage you or making promises which he ultimately doesn’t have the bandwidth to deliver. This frustrates you so you either deflect or ignore it causing him to become even more reactive. Everyone’s acting in good faith but the two of you stop being able to trust each other effectively.

Consequently, you start trying to force alignment by “showing rather than telling”, by “letting the results speak for themselves” to deliver value. This ruffles feathers within the org, but you have enough credibility and a long enough leash to pull it off. The results seem magical when you’re right, and lots of people get really excited and ask you to continue. Unfortunately, you find it hard to disentangle yourself from the project when you realize that the results aren’t what you intended, precisely because everyone’s now excited about these magical results. On the other hand, the same foresight that lets you spot broader foolishness gives you early warning when one of your projects is likely to fail. You see this before anyone else does and course-correct accordingly. But then you get labelled as flaky because you see this before anyone else is able to. Moreover, it’s hard for you to get help navigating any of this because no one deeply understands why you’re doing any of this.

This isolation compounds your frustration and loneliness and sets the stage for a vicious feedback loop. It’s hard to bridge the gap to becoming understood. So you lean deeper into “show rather than tell,” but that makes you even more illegible.

You’d love nothing more than to be seen and understood but you just don’t know how. You know that you’ve got the potential to create far more value and beauty within the company you’re already in. But you don’t know how to get your insights and prescriptions to land.

This cycle eventually culminates into the ultimate dilemma - should you stay or leave? Staying at the company is painful because the constant misunderstandings are extremely draining. Leaving is painful because you love the people there. You’ve got a lot of friends there, love your management chain and are otherwise happy. The money doesn’t hurt either. So walking away feels like you’re giving up on something amazing.

Neither choice feels acceptable, but putting it off amplifies the vicious cycle you’re in along with your stress.

Fortunately, this is a false dilemma. Both choices assume that the problem is entirely the situation (i.e. the job, manager, company, etc). The problem isn’t entirely “out there”. Rather, this dilemma and the vicious feedback loop are downstream of your inability to enroll your co-workers into changing their behavior and how they see the world.

You can’t control how others participate in this feedback loop, but you can change your participation. The prerequisite for inviting people to meet you where you are, is to first meet them where they currently are.

Every habit within an individual arises due to some self-protective or self-serving reason. Consider why you look both ways before crossing the street, shake hands with people when you first meet them, or avoid saying things that will piss off your boss. Each habitual pattern arises consciously or unconsciously due to some contextual reasons. Habits can and do fall out of relevance. For example, the same habits you built to navigate your last boss’s behavior may no longer be relevant with some new boss. Similarly, a scarcity mindset can help in an environment with genuine scarcity, but can be less helpful in an environment with more abundance. Inviting someone to change a habit invites them to ask whether the habit is still relevant. This question invites them to face the archetypical dilemma of psychological development - should I change or should I stay the same? Changing is scary for many reasons. This new habit may not “work”, your identity and therefore relationships within your environment have have to shift, or you could get hurt in the process. Staying the same once you’re confronted with the question is also scary because you could be wrong about staying the same, and could therefore get hurt in the process.

Facing and unpacking a developmental dilemma often requires substantial cognitive and metabolic resources. The deepest dilemmas are often repressed into the unconscious precisely because the individual doesn’t yet have the resources to face them. They might even know how to face them. Therefore, forcing someone to face a deep developmental dilemma produces a proportionate amount of fear.

Consider all those “foolish” company processes you point out to coworkers that just don’t “get it”. These processes are ultimately implemented and sustained by individuals engaging in habitual behavior. These habitual behaviors arose due to self-protective or self-serving reasons. Offering a deep insight that points out the inadequacy of such processes without an adequate container of trust and commitment forces people to face their own developmental dilemmas. That’s scary for them. A cascading loop of fear is created as each subsequent explanation for this deep insight triggers yet another transformational dilemma. Eventually, you get labelled as “confusing” or “crazy” because it’s more self-protective to face the dilemmas you trigger.

Your woes of being misunderstood are downstream of accidentally constantly tripping onto people’s developmental dilemmas.

Your relationship with your manager is not immune from this dynamic. He might not understand what you’re pointing at if it’s nested in dilemmas he can’t immediately access. But he’s formally accountable for you so he often experiences these dilemmas with far higher stakes. Inevitably, he copes by reaching for excessive control to create safety, which can start its own feedback loop if he experiences over-care for you. Eventually, you’re both in mutual threat response due to your reactive cycles. Even meta-conversations about topics like career goals might trigger any number of developmental dilemmas within your manager if these conversations challenge his sense of identity. For example, “If I admit that I can’t help him get what he wants, am I a bad manager?”

Your work is to practice the skill of enrolling people to change their habitual behavior to overcome their developmental dilemmas.

Developing this skill requires both internal and external work. External work changes your actions and internal work changes your being. Your internal work is to integrate various shadows in your unconscious that may stymie your ability to effectively enroll someone. For example, your need to demonstrate intelligence might override your desire to help. Or you might habitually dismiss pushback without genuine consideration. Or get really frustrated and impatient when others can’t “keep up” with you. In each case, your shadows stymie your attempts at enrollment by making it more about proving your worth rather than providing value.

Most people directly jump into selling a prescription to enroll a change in behavior. This only works if the recipient already knows what the problem is, and trusts you enough to accept your prescription. But they’re unlikely to recognize that they have a problem if you’re advocating for a change to a habitual, possibly unconscious self-protective behavior. Or perhaps they’re too scared to act on the prescription with you, if they already know they have a problem.

Effective enrollment involves successfully selling the diagnosis before selling the prescription.

Selling the diagnosis first builds trust. It shows you know what you’re talking about and it helps them feel seen.

The persuasiveness of a prescription depends on the persuasiveness of the diagnosis.

For example, I’ll likely just ignore you if you start insisting I eat broccoli to get healthier. However, if you convince me that I have diabetes and I’ll die within the year unless I eat better, I’ll likely start shoving those greens down my mouth.

The deepest developmental diagnoses generate more fear than shallower ones. The greater the fear, the more trust is necessary for them to even hear the diagnosis. They then need even more commitment to act on its prescription. On the flip side, each successful diagnosis generates more trust in the relationship. That trust, in turn, affords opportunities to create more commitment to enroll them in the corresponding change. Getting to the other side of a prescription then creates the ground to sell them on a scarier diagnosis. A skillful and measured interplay of diagnosis and prescription allows you to gracefully walk up a stair-step of trust and commitment. It feels “slower” in the short term but is often “faster” for landing durable change.

Enrolling others to undergo change is a deep topic. One essay can’t possibly hope do it justice. However, my coach has taught me an overarching sequence that’s been effective for me. Each step in the sequence sets up the subsequent one.

1. Name their unmet need.

   • This helps them feel seen. It lowers their defenses enough to hear what comes next.

   • Warning: Jumping straight to your diagnosis of why their need is unmet signals you care more about your insight than their experience. You’ll trigger a bunch of fear with nowhere for it to go, so they’ll write you off.

2. Describe visible symptoms of the unmet need.

   • Each symptom should err on the side of being “observable objective reality” that any reasonable person within the context would find hard to argue with. Your goal is to describe reality, not interpret it.

   • Warning: Rushing to explain the cause of each symptom signals you care more about demonstrating what you’ve figured out than helping them feel understood.

3. Paint their dream.

   • This shows you know what they want beyond merely pointing at what’s broken. This earns their permission to name the dilemma that’s keeping them stuck.

   • Warning: Skipping this might suggest you’re more interested in the problem than in them.

4. Name the dilemma keeping them stuck.

   • Naming and describing the competing purposes keeping them stuck should create both fear and excitement in their body. Excitement because they’ll start hoping that this is their opportunity to get unstuck. Fear because they’ll know that this is an invitation for change.

   • Warning: You can’t really proceed until you’ve landed this step well. Articulating a dilemma they don’t recognize signals you’re more interested in solving your puzzle than demonstrating that you deeply see the pain of their stuckness.

5. Deliver the diagnosis.

   • Skillfully laying out the deeper pattern underneath the dilemma should produce an “aha!” moment that simultaneously addresses fear and builds trust.

   • Warning: This step is load-bearing for everything that follows because it consolidates the trust that allows them to hear your prescription.

6. Land the key distinction

   • “Confusion” comes from the Latin verb “confundere”, which means “to pour together”. The boundaries of things “poured together” get dissolved, leading to foolishness and disorder.

   • Every developmental dilemma is downstream of some deeper confusion where multiple ways of being have become unhelpfully mixed up. The key distinction separates this confused mixing to make the prescription obvious.

   • Warning: A distinction that doesn’t click means you’re teaching your framework rather than resolving their confusion.

7. Sell the prescription.

   • The prescription should feel inevitable and unsurprising if you’ve done your job up to this point correctly.

   • Each successful delivery builds enough trust that they’re now willing to commit to the change you’re pointing at.

   • Warning: A prescription that surprises them often means you’ve skipped some important steps.

Steps 1-4 meet them where they are. Step 5 sells the diagnosis. Steps 6-7 sell the prescription.

The point of this essay isn’t to convince you to stay at your job or leave it. It’s to show you that your woes stem from your inability to enroll people within your current environment. Staying at this company may be a bad idea if the environment isn’t conducive to practicing enrollment. Similarly, leaving may not fix the problem if you leave and then don’t improve at enrollment.

Getting better at enrollment is the highest-impact path for you to feel seen, understood and to have your insights land for other people. Enrollment is a skill and therefore requires practice. Success won’t be overnight, but you’ll know you’re on the right track when you’re increasingly able to influence people over whom you have no formal authority.

Acknowledgements

Thank you to Brian Whetten for teaching me everything I’ve articulated in this essay.

Each essay in this series is derived from a composite of people, situations and patterns I’ve noticed during my decade at Google and across my life. Each essay attempts to capture the through-line going through such patterns in a way that clarifies a diagnosis and offers a specific prescription. It’s impossible to offer diagnoses and prescriptions that capture the sum totality of the interaction, and so these essays don’t even try. Rather, they’re an attempt at providing a developmental pathway for the protagonist of each story to grow and develop. Eventually, I’ll start providing examples of how such developmental pathways could be accelerated with AI.

This essay is for you if you manage someone brilliant who exhausts you. His passion and drive allows him to produce magic no one else on the team can. However, you find him to be deeply illegible and therefore unpredictable. You want to help but you’re stuck. It’s especially for you if the dynamics below match your experience.

Your report’s reasons for why he does what he does are totally illegible. His career goals don’t make any sense and he claims he doesn’t care about promotion even though all comparable peers are getting rapidly promoted. Probing deeper, he often says he finds the promotion process tiresome yet claims he only wants to work on the most “interesting” problems. He has the ability to amplify the energy of his collaborators when he becomes deeply excited about something. Unfortunately, he’ll often unpredictably lose interest and move onto something else before properly landing his first thing. His actions become somewhat clear in retrospect if his efforts succeed, but you’re often left holding the bag if they don’t.

You’d like to understand and help him. But his pace of work constantly outstrips the bandwidth you have. You’re already so oversubscribed across so many projects and other reports that you often can’t deliver on your commitments to him. So he subsequently decreases the frequency with which he keeps you in the loop. But then you start falling even further behind in understanding what he’s doing and why.

This growing opacity increasingly triggers your anxiety. Sitting still while he’s “running amok” feels negligent and intolerable, so you ask him more questions and check in more often. You try to generally supervise him more closely, but each interaction is now layered with ever increasing anxiety. He perceives this sudden increase in anxious contact as a form of micromanagement which triggers and increases his own anxiety and defensiveness. For example, perhaps he starts increasingly over-explaining everything he’s doing but these voluminous and energetic explanations make him even more illegible to you. Moreover, if he doesn’t feel heard, he’ll start sharing these insights with many other senior people within the company which spikes your anxiety even further. The more you push, the more he digs in. The more he digs in, the more you feel compelled to push.

1:1 meetings feel increasingly unproductive as everything happening in the backdrop escalates. Unlike with your other reports, your meetings with him generally function to feed your respective anxieties. He routinely offers diagnoses and justifications for his work, for the state of your org, the overall company and ecosystem that you instinctively want to disagree with. They’re often insightful if you stew on them for a bit. However, his prescriptions often feel scary, wildly unrealistic or otherwise out of touch with reality.

It’d be far easier if he were obviously an asshole or a low performer. Then the conversation would be about firing him. Unfortunately, it’s obvious that his actions aren’t malicious and that he’s broadly respected within the company. You can’t fire him and you can’t ignore him, so you feel trapped.

You’d like to build a connection with him to grow a relationship that doesn’t constantly spike your anxiety. Unfortunately, over time, you’re torn between two choices that each feel scary. You could potentially back off and give him more space. But if you do, you’re scared that he’ll eventually get frustrated that his efforts aren’t getting him promoted, blame you and then leave. Eventually, you’ll get blamed by leadership for not retaining him. Or maybe you’d see yourself as a “bad manager” who abandoned your report when it was your job to insulate him from the broader “dysfunction” of the org if he doesn’t get promoted or recognized for the value he creates. On the other hand, you could attempt to get far more involved with his work. Perhaps you could ask far more questions about his work and decisions. But you don’t really have the bandwidth to do this “properly”. So he perceives all of this as “micromanagement” which makes him increasingly defensive. Going down this path long enough starts turning you into the controlling manager you never wanted to be.

Locking into either choice feels untenable. So you feel increasingly stressed and stuck the longer you put off either choice.

The first step is to recognize that you’re both locked in a feedback loop of drama and reactivity. This feedback loop is sustained by three interconnected dynamics.

First, your anxiety distorts your perception. Your emotional state has become attached to feeling in control of the situation by way of your report’s emotional state. Anxiety floods in when he shows distress, confusion or makes choices you don’t understand. This anxiety acts as a filter and colors his actions as being more threatening than they actually are. In this flooded state you respond to this distorted picture of reality rather than what’s actually happening.

Second, beneath the anxiety are needs you haven’t owned or articulated. For example, perhaps you have needs like certainty, feeling competent, not being blamed by leadership. You can’t articulate these needs to him cleanly. So you try to meet them indirectly through control rather than making clean requests that your report could negotiate on. Eventually, control becomes your proxy for safety.

Third, beneath the unmet needs are insecurities he triggers. For example, perhaps you fear failure, being exposed as inadequate or feel guilt about not doing enough. These un-integrated shadows drive the anxiety that distort your perception and actions.

This dynamic works both ways. Your report also likely has needs he struggles to articulate without triggering you. You’re both unable to create clean agreements to get your respective needs met when you’re stuck in this reactive feedback loop. Neither of you can hold space for your own emotions and the other’s simultaneously. But notice that he’s a trigger for your reactivity rather than its cause. You can’t control his participation in this feedback loop, but you can change yourself.

Your authority as a manager gives you more levers to shift the dynamics of this relationship. It’s worth mentioning that you also have a greater responsibility to do so since you’re the leader in this relationship. Your work is to shift from a state of being in over-care to a state of caring.

Over-care might look superficially helpful but it’s caring distorted by your anxiety, unmet needs and shadows. It’s actually an attempt at managing your discomfort via control. Caring involves helping people by holding space for them to own their developmental work, without your anxiety contaminating the interaction.

Shifting from over-care to care involves both internal and external work. External work involves changing your actions (e.g. behaviors, practices and habits). Internal work changes your being (e.g. noticing, accepting and healing shadows). External work without internal work is limiting because it’ll quickly get undermined by your unhealed internal resistances. Internal work without external work is limiting because you won’t build the external skills necessary to actually change your situation.

Internal resistances are deeply contextual so it’s difficult to offer concrete prescriptions via an essay format. However, I’ll create subsequent posts to attempt exactly that. Internal resistances are best explored with the help of a therapist or executive coach. However, recognizing some of these resistances may nevertheless help. They include a desire for certainty, impatience, impulsivity to act, a fear of failure and blame, fear of job loss, needing to appear in control of leadership and guilt/shame about not doing enough or being enough. Your internal resistances map to the shadows that drive this entire feedback loop.

External work in contrast is far easier to describe and specify. The steps below illustrate the actions you can take to gradually shift from over-care to care.

1. Set an intention before every 1:1 to shift from “I am responsible for protecting him and myself” to “I’d like to better understand and empower him, and hold healthy agreements with him”.

2. Practice cultivating your presence within the conversation. For example:

   1. Pause before acting or responding to anything he says. Specifically, pause immediately when you feel the impulse to intervene, correct or control. Try waiting for one whole breath. This pause creates space for your over-care to get interrupted and reprogrammed.

   2. Use invitations rather than insertions. “Let me help you with that.” or “I’ll take care of this for you.” inserts yourself into his situation to soothe your need for control, but “Would you like help with that?” invites him to grow his agency.

   3. Use any instances of surprise at his behavior as fuel for developing mutual understanding rather than as an opportunity to correct or control.

      • “You’re going to do $BLAH?? I don’t think $BLAH is going to work. $BAD_THING will happen if you do $BLAH” demonstrates an immediate judgement despite you most likely lacking sufficient context.

      • “I’m curious about $BLAH. I think I can see how succeeding at $BLAH would lead to $BENEFIT. But have you considered $BAD_THING as a consequence?” demonstrates curiosity and creates an opportunity for shared understanding.

   4. If an invitation to participate from your report isn’t present, ask yourself what specifically you’d change before inserting yourself into the situation. Use the Bad/Good/Better/Best framework to check whether things are genuinely below the “red line” to warrant inserting yourself without an invitation. Moreover, prepare communication for your report to explain why you believe the situation to be below the red line.

3. His every answer whenever you ask him “why?” is an opportunity to deepen your understanding of what’s actually animating him. Reflect back what you’ve understood until he feels seen and heard. If his “why?” is aligned with yours, vulnerably articulate what you need from him as he progresses in his work. Ask him to reflect back your needs so that you feel heard.

4. Co-create explicit and specific agreements between the two of you once sufficient mutual understanding has been reached. For example, suppose he says he doesn’t care about promotion in the next cycle. Ask him to write a paragraph about what his career goals are, what broader purpose animates those goals, and what trade-offs he sees in achieving that purpose via those goals. This document can act as a formal agreement, and can be used as a tool for generating clarity when either one of you deviates from the principles and behavior articulated in the agreement. Such agreements can become anchors when the relationship starts growing in reactivity.

Shifting from over-care to care isn’t an outcome but rather an ongoing practice cultivated by the sort of habits articulated above. Success won’t be overnight but you should gradually feel less anxious in your 1:1s and more capable of holding space. Over time, you should see deeper alignment with your report while he grows increasingly empowered.

The next essay will explore this dynamic from the report’s perspective. It’ll explore what he can do to improve his participation in it.

Acknowledgements

Thank you to Brian Whetten for teaching me everything I’ve articulated in this essay.

It’s for you if you’ve accepted at least at a high-level that software engineering has changed forever, but haven’t yet figured out how to transform your team’s workflows. Perhaps you’ve used Cursor here and there, but whatever you’ve tried doesn’t seem to work for complex brownfield codebase problems. You know it’s possible to do these more complex things because you see other leaders/companies making it work. However, investing resources into this stuff is hard because the non-deterministic nature of LLMs introduces irreducible unpredictability and research overhead. This is compounded by the fact that you and your team feel like they’re constantly drowning to get the bare minimum done.

Consequently, it’s very natural for you or your team to feel stressed or defensive whenever anyone asks why you’re not using more AI to create value. In an ideal world, your team would have the ability to rapidly integrate each new AI innovation to create value proportional to the underlying improvements of LLM performance. But your team is trapped in a dilemma:

• Dive into AI (i.e. exploration) - Sounds nice, but you’re already underwater. There’s a risk you’ll do these experiments at the cost of “safer” work, the experiments won’t pan out and you’ll have nothing to show for it.

• Keep doing what you’re doing (i.e. exploitation) - AI is rapidly changing every quarter, so sticking with “safer” work also feels unsafe because you might suddenly find yourself irrelevant.

Locking into either choice feels stressful but so does deferring the choice, and this stress seems to grow every week as AI improves.

So what should you do?

First, notice that framing this dilemma as exploration vs exploitation misses the mark. In fact, there’s no “safe” choice available because both options tap into the same underlying scarcity mindset.

Nothing ever feels “enough” within a scarcity mindset. Exploration feels like theft from “real” work because it might not pan out, and exploitation feels like falling behind because the external world is changing so quickly.

This scarcity mindset then creates a vicious cycle:

1. Your sense of what’s truly “enough” isn’t sufficiently grounded in reality because fear, judgement and pain from your scarcity mindset distort your assessments. Perhaps each task must be done “perfectly” or you risk feeling like a failure. Perhaps there’s never enough time to do each task because everything feels urgent, despite the team’s overall prioritization.

2. This distortion of what’s enough amplifies your perception of scarcity.

3. This leads to you avoiding any “secondary” duties as a leader (e.g. talking to other leaders within the company to get context, running your own experiments with AI, etc.)

4. This avoidance increases the gap between overall organizational context and your assessment of it.

5. This increased gap makes it harder to understand what’s “enough” for yourself and your team.

6. Repeat step 1 and the vicious cycle repeats.

Building an organization that can fluidly absorb novel AI capabilities requires psychological space to experiment. This in turn requires clarity about what’s actually “enough”. So the deeper you fall into your scarcity mindset, the harder it is to create novel value with AI.

To be clear, possessing patterns around a scarcity mindset isn’t “bad” or “wrong”. They’re self-protective patterns that have genuine value within environments containing emotional or material scarcity. But it’s worth asking whether these patterns are still relevant, or whether they’re running on autopilot in your mind. Moreover, it’s unrealistic to flip a switch to transition from a scarcity to an abundance mindset precisely because they’re often so deep-seated.

Unlike the scarcity mindset, an abundance mindset is a different tone for relating to the world. It’s something like “I now have enough to start creating value for everyone else”. Where a scarcity mindset sees threat from change (e.g your company’s value chain, workflows, etc), an abundance mindset sees opportunity to create value.

Undergoing a shift from scarcity to abundance requires you simultaneously engage in internal and external work. External work involves changing your actions (e.g. behaviors, practices and habits). Internal work changes your being (e.g. noticing, accepting and healing self-protective patterns that generate scarcity).

External work without internal work is limiting because it’ll get quickly undermined by your un-healed internal resistances. Internal work without external work is limiting because you won’t build the external skills necessary to actually change your situation.

A core external practice for engaging in this shift is what my coach calls Bad/Good/Better/Best. This practice gradually improves your ability to track what’s enough, and creates space for you to increasingly create more value. Internal practices are far harder to describe given how contextual each person’s internal resistances are. I’d recommend seeing a therapist or executive coach to help with any resistances that arise as you attempt this practice.

Although this practice can be run as a team, it’s often easier to shift your team from scarcity to abundance if you yourself have made that shift first. The rest of this essay assumes that for now you’re the only one that’s working on this shift.

1. Forming tasks for the next planning cycle

Start by organizing your time around explicit planning cycles. You may initially keep these private as a tool for organizing your time. You may start with a one-week planning cycle given that you’re the CTO of an early-stage startup. It’s natural to experiment with shrinking or growing the duration of these cycles as your role’s volatility changes.

Next, start gathering tasks for yourself and classify each item according to the following:

• Bad (i.e. the red line): Getting everything done in this bucket is the true “minimum”. If it isn’t done, genuinely destructive things will happen. Completing these tasks keeps the lights on.

• Good: The actual value created beyond keeping the lights on.

• Better: Tasks that would create more value than Good.

• Best: The best-case outcome of the value you’d create within the planning cycle.

Tracking the red line is inherently nebulous because it depends on stakeholder moods, competitive dynamics, investor expectations, and other contextual factors that may shift weekly.

The skill of tracking the red line is the crux of an abundance mindset. It’s the mindset of forming an initial hypothesis of what’s enough, running experiments and course-correcting based on what you learn. Your approximations will improve over time with reflection.

Here are some internal distortions that may impair your ability to track the red line. All of these point to some sort of internal work that you may do with your therapist or executive coach:

• Anxiety about seeking clarity - You’re often worried about being seen as “weak” or “pessimistic” if you ask your boss hard questions about your team’s priorities and resourcing.

• Guilt about your work being merely “good enough” - If you’re someone that feels like failure if you consistently fail to exceed the “Good” standard.

• Absorbing others’ urgency - Stakeholders (especially people with more power than you) may regularly make polemical and emotional cases for why certain work should immediately get prioritized. Absorbing this urgency without reflection can be counter-productive.

• Getting dejected by a single failed AI workflow experiment - This is if you temporarily lose all motivation, self-confidence and willingness to experiment the moment any individual AI workflow experiment fails.

2. Reflecting on this cycle’s triage

Next examine your initial triage to inspect the distribution of Bad/Good/Better/Best tasks. Bad’s the bare minimum and should have the smallest number of tasks. Best should have the largest number of tasks. Deviances from this distribution aren’t “bad”. However, it may imply that either the organization is operating at capacity, you are operating at capacity or that you’re all deluding yourselves about what’s enough. The point isn’t to force yourself to fit some ideal distribution, but to use the distribution as a provocation to increase awareness.

It’s also worth noticing whether you can coherently articulate and justify your choice of where the red line is. If you can’t, it’s worth examining why not.

3. Maintaining a tasks backlog

You’re likely getting inundated with new tasks constantly. Many new tasks may create the competing pressure to either drop everything to react to it or to feel guilt and fear for ignoring it.

An alternative is to test each task against the red line. If they’re below the line, they get prioritized for action this cycle. Otherwise, they get added to a backlog for the next triage. It’s better to change the length of each planning cycle rather than breaking the integrity of the process.

Maintaining such a backlog is particularly relevant for filtering novel AI workflows from the “noise” on Twitter/Slack.

4. One right-sized AI experiment

It can be helpful to pre-specify a clear timebox for each experiment popped off the backlog, since LLMs are non-deterministic and may require an unpredictable amount of experimentation.

It’s crucial to prioritize experiments which if they succeed, you’ll use everyday. Over time, this will compound your capacity to create value with AI.

5. Celebrating both failures and successes as learning experiences

It’s worth taking a minute or two to celebrate the shift you’re making from scarcity to abundance irrespective of the outcome of each AI experiment. This gradually builds positive associations with an abundance mindset.

The shift from scarcity to abundance isn’t a one-time event. Rather, it’s a set of habits you develop augmented by the inner work of overcoming various resistances and shadows.

Acknowledgements

Thank you to Brian Whetten for teaching me everything I’ve articulated in this essay.

So you respond by pulling all the levers you know. Reorgs, new OKRs, AI adoption metrics and re-leveling compensation to light a fire under people’s asses. Yet six months later, the company’s aggregate outcomes haven’t changed in any meaningful way. It’s like your interventions got absorbed by something you can’t quite pin down.

“Culture” typically gets blamed when such transformations fail. But it’s a big word and there’s too much fuzziness in how that word gets used. Moreover, it’s hard to prioritize something you can’t define when there’s substantial pressure to improve profit.

Each leader responds to this pressure in one of two ways. Each example has been deliberately exaggerated and caricatured to make the broader point.

• Mechanistic view: You create new roles, reporting lines and compensation tied to extremely crisp AI outcomes and light a fire under everyone’s ass. A year later, people thoroughly gamed and Goodhart’ed the metrics in unforeseen ways leaving the company just as politically stuck as before. In the mechanistic view, culture is an aggregation of incentive structure designed to drive specific outcomes.

• Mystical view: You feel it’s authoritarian to dictate what people should value in a top-down fashion. You want everyone to feel “included”. So you organize lots of annual company gatherings, listening tours, etc. that are full of deeply moving rituals and ceremonies. A year later, your most politically astute and rapacious VP has filled the power vacuum, your best people are leaving and nothing is getting done. The mystical view sees culture as an ineffable and intangible thing that emerges organically and can only be participated in but never directed.

Each view is tempting for leaders because each contains a kernel of truth. Culture indeed involves stated values, incentive structures and emergent properties. But none explains why interventions keep landing differently than intended. You know that culture matters, but you also need results now. You’re torn because you feel that investing in one means neglecting the other. In the meantime, your AI pilots keep failing and the company keeps chugging along exactly as it did. Your stress levels rise in tandem as a result.

Getting unstuck from this pattern requires understanding what culture actually is and how it’s causally connected to your company’s value chain.

What is a company’s culture?

Companies attempt to organize employee behavior via constraints (e.g. rewards, punishments, rights and responsibilities, rituals, physical spaces, etc.) to maximize profit. These constraints can be intentionally designed or unintentionally emergent.

But constraints are only half the story. What people do based on such constraints depends on how they interpret them. For example, the same incentive can be read by one employee as genuine commitment toward growth, and by another as cynical manipulation. The same feedback ritual can feel like development or repressive micromanagement. The same promotion can signal “we value excellence and meritocracy” or “we reward political ambition.”

The constraints themselves (e.g. rewards, punishments, rights, responsibilities, etc) are tangible and clearly patterned and live in objective reality. The interpretations of these constraints are nebulous subjective experiences living inside each participant’s mind, either consciously or unconsciously. Having said this, there are limitations on not just the physical constraints that are possible within a given context, but also their corresponding set of subjective interpretations. For example, you can’t offer rewards with money you don’t have. Similarly, few employees in the US would interpret a public flogging and shaming as a reward.

Constraints and their interpretations might seem like separate things, but they’re actually inseparable. Neither constraints nor their interpretations are causally primary, and are in a feedback loop with each other. For example, as a leader you may decide to impose specific constraints based on shifting organizational objectives. You can’t read people’s minds and you’re not omniscient. Therefore, you’ll need to make some implicit assumptions about how your constraints will be interpreted and therefore how behavior will change. But given that you’re not omniscient, you’ll at least somewhat miss the mark. There will necessarily be a gap between how you expect the constraint to be interpreted and how it’s actually interpreted. This interpretation gap lies at the heart of most cultural difficulties.

Culture is the ongoing feedback loop between the interventions a group makes on itself and the interpretations those interventions produce.

This feedback loop is inevitable. It emerges whenever multiple people engage in organized, committed relationships of value creation across time. A company’s cultural health is a function of the robustness with which its overall interpretation gaps are identified and navigated towards the pursuit of creating value. This is easier said than done.

The difficulty in closing this interpretation gap explains why acting narrowly from the mechanistic or mystical views fails in practice. The mechanistic view focuses on constraints and incorrectly assumes that interpretation of them is uniform within the company. The mystical view focuses on interpretation and incorrectly assumes that any sort of constraints are tyrannical. Each view misses that culture is the feedback loop itself.

Why is this interpretation gap so hard to close?

Closing the interpretation gap requires seeing how your perception of reality contrasts with the feedback you receive from it. But your ability to navigate the nebulosity of such conflicting information is limited by your own psychological development. Therefore, companies that prioritize closing their overall interpretation gaps must prioritize the development of their employees, which is hard.

Why is psychological development hard?

Jean Piaget was a developmental psychologist. He was intrigued by the fact that children of different ages made different kinds of mistakes. These weren’t random errors or less accurate versions of adult thinking. They were systematic errors downstream of sophisticated reasoning limited by the child’s cognitive stage.

For example, Piaget studied children under the conservation of number task. It involves showing a child two rows of tokens in one-to-one correspondence. After the child confirms they’re equal, the tokens are spread apart and the child is asked again if they’re equal. Children below the age of seven reliably say the longer row has more tokens, even though nothing was added and removed. This isn’t carelessness. It’s a coherent but flawed reasoning strategy. The child focuses on one visual dimension (i.e. length of the row), while ignoring another (i.e. how bunched up the tokens are). The meaning-making system within children at this stage of development engages in systematic errors that aren’t present in older children. Moreover, each stage of development in both children and adults is characterized by different families of systematic errors.

Both children and adults experience internal cognitive conflict when our existing mental maps encounter experiences they can’t adequately assimilate. This mismatch can often be deeply uncomfortable, especially if it’s large. On the other hand, it motivates accommodation and drives psychological growth.

Development isn’t about acquiring more knowledge. It’s about overcoming systematic errors in how you make meaning.

The Kegan stages of development articulate how adults grow capacity to navigate increasingly complex spaces of possibility. Each stage can be viewed more crisply in terms of the systematic errors it makes:

• Stage 3 (Socialized Mind): You get stuck when different social groups want mutually exclusive things. For example, if your team wants one thing but your boss wants another. So at this stage, you’ll struggle to notice interpretation gaps that require holding conflicting social expectations simultaneously. Especially if each expectation seems legitimate.

• Stage 4 (Self-Authoring Mind): You can overcome these competing values by constructing your own self-authored values. You can say “I hear what both groups want, but here’s what I believe is right”. But this self-authoring makes you vulnerable to rigidity since it can be the answer to everything. Even when the situation calls for something entirely different. For example, if you deeply value “radical transparency” and your new performance management system lands as surveillance, your framework may not let you easily see this.

• Stage 5 (Self-Transforming Mind): You can overcome this rigidity by participating within multiple perspectives simultaneously. You relate to competing values as an opportunity for collective growth and value creation, rather than something to be neatly solved. You can simultaneously optimize for these competing values without losing overall coherence.

Development doesn’t just change one’s objective behavior. It changes one’s subjective experience and how one makes meaning. That is, it changes the depth of nebulosity one can navigate, and therefore the complexity of patterns one can navigate. For example, what looks like a personal attack to a Stage 3 looks like useful feedback to a Stage 5. This process is simultaneously emotionally taxing and rewarding. So it requires participation within conducive containers to provide appropriate curriculums and support.

The absence of sufficient development within a company’s leaders makes closing the interpretation gap hard. For example, consider a time when lots of people tried to consistently give you the same feedback, but it was only after a “come to Jesus” moment that you finally “got it”. That’s exactly the phenomenon we’re pointing to. Until your “come to Jesus” moment, you lacked the capacity to integrate this feedback due to your own systematic errors of meaning-making, and therefore the capacity for that leader’s organization to navigate the interpretation gap is fundamentally limited.

Why do the two views of culture fail?

The mechanistic view assumes that interpretation is uniform or predictable. Design the right incentives, get the right behavior. This is the systematic error of treating humans as input-output machines. The same incentive structure will be gamed differently depending on what people believe about the company’s actual priorities. Someone who interprets the company as fundamentally political will optimize the metrics differently than someone who interprets it as fundamentally meritocratic.

The mystical view correctly notices that interpretation matters but concludes that culture therefore can’t be shaped. Or that it’s tyrannical to attempt to do so. This is the systematic error of confusing nebulosity with intractability. Culture is tractable but not mechanical.

Moreover, both views miss the feedback loop, and that culture isn’t purely the constraints nor their interpretations. Rather, it’s the ongoing dynamic process between them.

What is the relationship between culture and profit?

Viewing the company’s culture as a feedback loop between constraints and their interpretations gives the space we need to clearly see the connection between culture and profit. Specifically, that they’re deeply interpenetrating and two sides of the same coin.

But first, let’s concretely articulate what a company is in a manner conducive to answering this question.

What is a company?

A company is an organization of people in a specific and evolving pattern of interaction, organized around creating some external value in the market.

Michael Porter‘s work on competition provides the concepts we need to crack this open. In his conception, a company’s value chain is the specific sequence of activities (e.g. design, production, marketing, delivery, support, etc) arising from employee interactions that transform the company’s inputs into valuable outputs that customers will pay for. A company can achieve competitive advantage (i.e. above-average profits within a market) by competing to be different, rather than competing to be “the best”. Specifically, by differentiating along its value chain to fundamentally think, do and be different from the competition, rather than playing the competition’s game marginally better. See this earlier post for a review of his work.

A company’s value chain is a pattern of activities within objective reality. But as discussed, a company also includes a coherence of shared interpretation and aligned subjective experience. A company’s employees develop common ways of making sense of events. Whether something counts as “success” or “failure”, or whether a given incident is salient towards value creation or simply background noise. These shared interpretations evolve in feedback together with the value chain. They co-create the company’s market outcomes.

Founders starting a company each bring their own developmental challenges to the mix. At conception, their systematic errors in meaning-making shape how the entire company interprets events. This gives way to the company’s aggregated psychology characterized by the systematic errors the company as a whole engages in.

How does a company’s aggregated psychology emerge?

Priya and Marcus are co-founding a B2B SaaS startup.

Marcus tends toward Stage 3 patterns so he interprets customer complaints as indictments of his worth. When a big client threatens to churn, he either freezes or over-accommodates because he’s unable to cleanly separate “our product needs work” from “I’m a failure”.

Priya tends towards Stage 4 patterns so she has a tendency to believe that her self-authored framework is the answer to everything even when it isn’t. Additionally, she has strong opinions about product direction and interprets pushback on her vision as an overall lack of understanding.

These developmental patterns interact and build upon each other within their company. Marcus tends to defer to Priya when he feels overwhelmed by criticism because he finds her certainty stabilizing. On the other hand, her rigidity goes unchecked because Marcus can’t push back without feeling like he’s attacking her personally. So customer feedback gets filtered through Priya’s framework. Additionally, investors and advisors internalize that raising concern about product direction creates tension with Priya and distress for Marcus. So their overall dynamic creates a norm where problems get minimized until they become crises.

This isn’t inevitable. Different patterns could emerge if Priya and Marcus worked on their developmental edges, perhaps in executive coaching, therapy or mutual support. Priya could learn to see her framework as one perspective among many. Marcus could learn to separate product feedback from self-worth. This would create space for earlier, calmer course-corrections.

Similarly, more people joining the company leads to the emergence of an aggregated psychology that’s a function of the developmental trajectories of each employee proportionate to how much power they wield within the company. All of these systematic errors either cover for each other, or compound on top of each other. Like individuals, companies develop aggregate patterns of systematic errors which grow more complex as more people join.

How does a company’s culture and value chain co-create profit?

The specific pattern of interactions between Priya and Marcus give rise to a specific value chain that produces profit. Their idiosyncratic preferences, values and tastes shape the space of possibility in terms of the constraints they might apply to their two-person company, and how they may each interpret these constraints. But on the whole, their goal is to organize and optimize their behavior so as to increase the profit generated by the company.

This increase in profitability in turn places reciprocal demands upon the company to complexify its operations. For example, perhaps there’s a lot more market demand than their infra or ops can currently handle. This combined with their currently limited resources as a two-person company forces them to make more sophisticated trade-offs within the value chain. The space of viable trade-offs is obscured by the company’s systematic errors (i.e. psychological development).

Nevertheless, these initial trade-offs allow them to cope with increased demand and start to differentiate their value chain. So profits continue to increase.

As profits increase, this overall feedback loop starts to spin:

1. Increased profits and demand leads to more employees getting hired to meet demand.

2. More employees leads to more constraints getting imposed by leadership to organize this growing headcount towards profit.

3. There’s an interpretation gap between these constraints and their interpretation.

4. Each employee’s interpretation is a function of their specific psychological development.

5. Their interpretation of the constraints causally influences the tactical and strategic decisions that they make to meet this increased demand. Such decisions get made concurrently across the company.

6. Such decisions imply that trade-offs get made within the value chain that are idiosyncratic to that company’s aggregate psychology. Some of these trade-offs are desirable whereas others aren’t.

7. Nevertheless, these trade-offs allow the company to meet increased demand and generate more profit.

8. Leaders then hire more employees, and the rest of the loop begins to spin.

The idiosyncratic trade-offs within the value chain are the joint point between a company’s culture and value chain. The company’s competitive advantage is therefore constrained by how developed the aggregate psychology of the company is. Certainly, there are many other factors that influence a company’s profitability. But the differentiation of a company’s value chain is inextricably linked to the company’s culture. Profit and culture may appear as competing considerations but they’re in some sense two sides of the same coin.

Founders who think culture is a luxury that they’ll attend to later are already shaping culture through their systematic errors. The question isn’t whether one’s deciding to shape culture or not. It’s whether one’s willing to do so consciously and skillfully.

Good culture develops its people

I’ve deliberately used the term “leader” rather than “manager” throughout this essay. In my model, if someone consistently does what you tell them you’re their leader within that context irrespective of whether they formally report to you. Formal authority is a constraint to backstop against fluid authority degenerating or stymieing value creation. Anyone within a company can grow their power and leadership.

As my executive coach says, leadership is the act of taking accountability for creating value with integrity.

A company has many more leaders than the executives. The aggregate capacity of all these leaders in closing their own interpretation gap determines the company’s overall ability to navigate the constraint/interpretation feedback loop. A company’s culture can either entrench people’s developmental limits or help them grow past them.

A culture that develops its people surfaces systematic errors at a pace people can metabolize rather than entrenching them. A good culture affords a feedback loop:

1. Developing people allows each of them to commit better systematic errors.

2. This improves their ability to close the interpretation gap.

3. This improves the chance that their interventions will be better integrated.

4. This improved integration affords more capacity to further develop its people, and so on.

Conclusion

Culture isn’t a nice-to-have you attend to once the business is stable. It’s the ongoing feedback loop between the constraints you impose and how those constraints get interpreted. Moreover, a company’s culture and value chain are interpenetrating. These loops are inevitable.

This doesn’t mean culture work is easy or that it guarantees results. But it does mean you can stop experiencing culture and performance as competing priorities demanding different resources. It’s as if they’re two sides of the same coin.

Future essays will explore how organizations can deliberately attune their culture toward development.

Acknowledgements

Brian Whetten, Prof John Vervaeke, Charlie Awbury, David Chapman for everything they’ve taught me.

Dan Hunt for helpful discussions, editing and generally co-creating these essays with me.

Yet it’s fairly easy to rapidly generate mediocre outputs that are “good enough”. This can be particularly deflating to all of you strong writers within an organization (e.g. PMs, Engineering Managers, UX Designers, etc) who approach your craft with a deep sense of aesthetics and caring.

This new world puts you between a rock and a hard place. On the one hand, you feel that if you don’t adopt AI you’d get left behind. Perhaps the PRDs or performance reviews that used to take you hours are being done by someone else within minutes. They’re clearly mediocre compared to your work but are “good enough” within your company’s culture. Moreover, you don’t yet trust AI or your skills in wielding AI to produce outputs you’d consider actually good. So it’s incredibly tempting to defer using it. On the other hand, you don’t want to produce the sloppy mediocrity yourself since it would challenge your deep sense of aesthetics and commitment to your craft.

Others, like you, caught in this tension will typically thrash and get endlessly stressed and frustrated until they give up. Eventually, they collapse into over-control or under-control.

It’s tempting to read into this frustration as a sign that you’re not suited for AI. Or that achieving these new standards of “productivity” necessarily requires surrendering your standards. Neither is true.

The discomfort is indeed very real. But I’d offer that there’s a happy middle between these extremes. One that can rapidly improve your LLM’s outputs without jeopardizing your productivity.

Why is it so hard to produce good outputs?

Natural language is inherently underspecified. There’s no natural language communication that fully captures the meaning of that sentence. Effective communication between two individuals relies on a shared set of implicit assumptions between those individuals.

Imagine reviewing a pull request from a brand-new hire and commenting only, “Clean it up and then I’ll approve it”. Without shared context, that instruction is wildly underspecified. You’d have to spell out what “clean” means in excruciating detail. Onboarding exists to build that shared set of assumptions so a vague comment like this becomes legible without explanation.

This onboarding heuristic can be helpful for working with LLMs. They’re expert role-players with access to all public knowledge that have no pre-existing relationship with you. Imagine that every time you submit a prompt a temp worker gets hired with a 200k context window. They have the sum of all human knowledge at their fingertips but absolutely no idea who you are. They’re oriented towards broad societal values but don’t know what your team values at high resolution. This is germane for highly subjective or context-sensitive tasks. It’s common for model outputs for such tasks to regress to the mean of the training distribution and produce slop.

Subjective tasks like creative writing are especially hard because nebulosity confronts you immediately at every layer of abstraction in a way that’s less true for tasks like code generation. Failure to sufficiently frame and communicate what you want rapidly leads to underspecification in the prompt and eventually slop.

Subjective tasks also lean far more on an embodied and intuitive sense of what’s “good”. You often can’t fully specify what you want because you don’t fully know until you see some concrete instances and see how you feel. So a frontier model’s propensity to generate slop isn’t just about missing “information”. It’s also about missing crucial details that specify the overall problem that doesn’t yet exist in an articulable form. You feel something is off before you can articulate why. This isn’t a bug. It’s the nature of nebulous concepts like “good writing” or “helpful feedback”.

LLM folk-wisdom has devised role-playing prompts (e.g. “Act as a software engineering manager at a FAANG-caliber company.”) in an attempt to provide more of this framing. Such instructions do move the model closer to the right spirit. A software engineering manager and a professional artist may approach the same software engineering task very differently. But “software engineering manager” is still a broad category. It doesn’t capture what your team values, your coding standards or your communication norms. Role-play narrows the space but often insufficiently so for many subjective tasks like writing.

Endless iteration on a prompt without a coherent strategy to navigate the nebulosity within the space of iteration often fails, because it treats a stochastic system as deterministic. But it’s common because creating value with LLMs is uncomfortable.

What can you do about it?

You shouldn’t simply attempt to specify everything up-front in your instructions (i.e. over-control) and neither should you meekly accept whatever the model produces the first time around (i.e. under-control).

There’s a better third choice. The technique below provides a structured process to iteratively surface and correct key implicit assumptions that actually matter for your subjective task. This technique is quite general and can also be useful for complex objectively verifiable tasks.

Here’s the high-level core loop as a list of steps.:

1. Add instructions to your prompt asking the model to surface its implicit assumptions before generating its output.

   1. Editing something is often far easier than generating something. LLMs can rapidly generate lots of text.

   2. Shifting the burden of generating these assumptions also shifts a substantial amount of onboarding effort onto the LLM. Your role shifts to iteratively critiquing the assumptions it generates.

2. Inspect the assumptions before evaluating the output. If any are wrong, add clarifying context to your prompt and re-run.

3. Once the assumptions look right, evaluate the output. If it’s still off, add instructions describing what “good” would look like and re-run.

4. Repeat until you’re satisfied or run out of patience.

The technique works because it creates a feedback loop between two capacities:

• Embodied discernment: Your intuitive sense that something is off in the output, even before you can articulate why.

• Specified Patterns: Your provisional, explicit sense of what “good” means, accumulated in your prompt.

Your embodied intuitive discernment notices something’s off. You take this nebulous feeling and articulate a concrete pattern that becomes a new pattern in the prompt. Articulating this pattern provides sharper language to capture what “good” intuitively means to you. Sharper language makes it easier to notice finer-grained distinctions the next iteration, which leads to more refined patterns, and so on.

Embodied discernment without the capacity to articulate concrete patterns remains vague (”I don’t like it but I can’t really say why”). Articulating concrete patterns without embodied discernment creates a tyranny of meaningless checklists. If integrated in feedback, the two spiral upwards together.

This approach allows you to rapidly “onboard” the model onto your culture (i.e. the specific values, framings, quality standards, etc.) within the scope of the context window to make its outputs feel like they’re yours. In this approach, prompt engineering is inherently iterative.

Detailed step-by-step explanation of the technique

Setup (one-time per task)

Step 1

Structure your prompt with three explicit sections:

• Why is this task valuable?

• What outcome do you want from the model?

• Any additional details on how you’d like it to be done.

• Any other informational context or assumptions that the model should make.

How much to include is an empirical question that’s deeply task-dependent. As a useful heuristic imagine a temp worker with 200k tokens of memory and access to all public knowledge. Would they find your description legible? They already know all public knowledge. So you wouldn’t waste time explaining basic concepts (e.g. profit and loss, how gravity works, etc.) to them. Prioritize knowledge that’s esoteric or idiosyncratic to your specific context.

This is also why the overall process works. Getting the model to surface assumptions lets you lazily bootstrap context from the minimal dose the model needs to produce “good enough” outputs for iteration, rather than trying to specify everything upfront.

Here’s an example of what such a prompt looks like:

## Why this matters

I’m trying to turn a messy brainstorm into something I can share with my editor. The outline needs to be legible to someone who wasn’t in my head when I wrote the original.

## What I want

Convert my stream of consciousness document into a coherent outline.

## How I want it done

Use nested bullets. Group related ideas. Preserve my phrasing where possible; don’t sanitize the language.

## Additional assumptions

I’m writing for a technical audience familiar with LLMs. The final essay will be ~1500 words.

Step 2

Add instructions asking the model to surface its implicit assumptions before generating its reply.

## Response Format

The definition of “success” on this task is deeply contextual. It involves making a substantial amount of implicit assumptions about a wide array of things. Especially since natural language is really underspecified.

It’s easy for me to forget what context you may or may not have.

Please start your reply with a section of nested bullets called `Implicit Assumptions`. The goal of this section is to make explicit any implicit assumptions you’ve made based on the instructions I’ve provided in this user turn. Correcting these implicit assumptions would allow the two of us to get onto the same page.

Each assumption should be concrete, specific and load-bearing.

- The assumptions should not be redundant with the explicit context and instructions I’ve already offered here. After all, they’re meant to be *implicit* assumptions that I’m asking you to make *explicit*.

- The assumptions should be load-bearing in the sense that if I invalidate any of these implicit assumptions, one would necessarily expect the rest of the reply to materially change.

Please group these implicit assumptions based on:

- Why I value the response.

-Concrete characteristics of a good response.

There’s room to experiment with these instructions. I’ve arbitrarily chosen the dimensions of “Why I value the response” and “Concrete characteristics of a good response”. But those may not be the most germane dimensions for your task. Choose dimensions that would help the LLM best onboard onto your task.

Evaluating Assumptions

Essentially, we first check that both the categories of assumptions and the assumptions themselves for this task seem correct/reasonable before we proceed to directly change the output.

Step 3: Run the prompt.

Step 4: Inspect the generated assumptions.

Step 5:

• Do the default categories (i.e. value, concrete characteristics) of the generated assumptions seem germane and helpful for this task?

• If not, consider how the categories could be improved, and try changing the original prompt. Then go back to step 3 to re-run it.

Step 6:

• Do the assumptions within the current set of categories seem correct for this task?

• If not, add a new section to the original prompt called ## Important Assumptions, and add bullet points of corrections in there. Then go back to step 3 and re-run it.

Evaluating Outputs

Step 7: Examine the output. Specifically, check if it feels “off” in any way. This is an example of embodied discernment.

Step 8:

• If the output is “off”, imagine a specific pattern of what the model should have done instead. Add an instruction for it as a bullet point to the original prompt in a new section called ## Important Instructions. This crystallizes another pattern for the overall specification of the task.

  • If you have trouble what

• Ensure this additional instruction is a positive instruction. Negative instructions of “don’t do X” are harder for models to follow than positive instructions (“do Y”). Most instructions to correct model mistakes can be written as positive instructions. It’s easiest to get what you want when you’re clear about what you actually want. For example:

  • Example 1

    • Bad - Your reply shouldn’t be too long.

    • Good - Your reply should be no more than five sentences.

  • Example 2

    • Bad - Don’t reply in full sentences and don’t be long-winded in your response.

    • Good - Your response should be entirely in nested bullets. It’s okay to sacrifice grammar for length. Ensure each bullet makes a direct and specific point.

Suppose I’ve spent a bunch of time tuning the prompt. How can I be certain that my current attempts are the best that I can do for this task?
You can’t. Welcome to the nebulosity of stochastic systems. As discussed in my previous post, fault attribution and iterating on complex systems is extremely challenging.

Eventually, the output will stop triggering a reflexive “this feels off” gut response. Trust that instinct. Working effectively with AI places greater demands on your capacity for self-trust and embodied discernment. Similarly, having an unchecked sense of perfectionism can be an impediment to using AI since certainty is structurally unavailable.

It’s possible that your specific task is too complex for the method in this post. It’s helpful to approach all such stochastic tasks with a timebox. This creates space for check-ins to decide whether to keep pushing on the current approach, get help, or try a more sophisticated approach.

Closing

It’s not possible to create a prompt tuning process that can generate certainty for most subjective tasks. Moreover, crafting a process to seek such certainty is a non-goal. The process described above certainly doesn’t provide any such guarantees. This isn’t a limitation. Patterns are provisional. Grip them too tightly and you get over-control. Surrender to the model and you get under-control. The middle path involves forming patterns provisionally, staying connected to embodied discernment, and revising as necessary within an iterative loop.

Cultivating this capacity involves deliberate practice. Certain tools can make the practice easier. Custom GPTs, Claude Projects, Gemini Gems, etc. let you save and reuse accumulated patterns. A future post will provide concrete examples on GitHub accompanying instructions.

The capacity to navigate nebulosity isn’t uniformly distributed. Within a given domain, some people are better at facing it than others. It also doesn’t seem to be distributed by IQ. I’ve met really intelligent people that were just awful at dealing with it, and unintelligent people that tolerated it quite skillfully. This is certainly anecdotal. But something interesting is going on here.

The word “development” evokes many meanings to the average tech bro. For example, professional development, software development, skill-building exercises, etc. That’s not what this essay is about. This essay is about development in the “developmental psychology” sense. It’s about the capacity to make meaning when the rules run out, the world becomes fuzzy and one’s confronted with a slate of difficult choices. It’s about one’s capacity to navigate increasingly complex ways of meaning that can coordinate more perspectives, over longer time horizons, with less distortion.

The capacity to navigate nebulosity is developmental. That is, one’s ability to navigate nebulosity without either rigidifying it into false certainty or dissolving it into nihilism, while still producing a workable pattern for action, is a function of one’s psychological development. This essay’s goal is to clearly point out that adult development exists, and to make its connection with the capacity for navigating nebulosity explicit.

This isn’t a personality trait. Neither is it something one can learn the definition of and immediately apply. It’s a way of making meaning that grows through specific kinds of experience and challenge. Some people have consciously or unconsciously cultivated more capacity than others. One’s psychological development shows up most clearly when they’re embedded in a situation that genuinely resists clean answers.

Everyone’s seen real-life examples of what I’m pointing towards. Why is it that some people seem more “mature” when confronted with a difficult situation, and others don’t? That difference isn’t random.

Development in children

I have a nephew that recently turned one. When he was a newborn, he lived in a small “box” (i.e. crib) in his parent’s apartment. Both his perception and actions were incredibly limited. From the perspective of an adult, his field of action was very concrete and patterned, and not very nebulous at all.

The “box” widened as he grew and learned to crawl and eventually walk. His world became more complex, uncertain and nebulous at each developmental transition. Each stage invited him to master new levels of emotional processing, bodily control and sense-making to cope with this nebulosity.

Pre-school will likely offer him rigid rules to help him understand what it means to be a “good boy”. For example, “Hands are for shaking, not for hitting. Legs are for walking, not for kicking”. Despite seeming concrete to an adult, such rule-sets can seem quite complex for the impulsive brain of a child.

The criteria for being a “good boy” grows increasingly complex with age. For example, what if someone tries to mug him when he’s 16? Or when he’s a legal adult? The previous rigidity from pre-school around never hitting or kicking would seem irrelevant. At that point, the nebulosity of being a “good boy” would invite questions of proportionate force, ethics, etc.

It’s culturally obvious to us that psychological development exists in children. There’s entire pop-culture books on it. Developmental psychology’s less well-known claim is that this process continues well into adulthood. The stages of adult development are less visible but no less consequential.

Development in adults

Kegan’s six stages of development

The developmental psychologist Robert Kegan has created a map of the various stages of adult development. Note that adult development is a topic of active research. It’s far from a settled science. Nevertheless, Kegan’s work on stage theory can be extremely useful in making sense of relational and organizational dynamics. It’s best to relate to this work from the spirit of “All models are wrong, but some are useful”. It’s been my experience that an individual’s participation within a stage may be deeply contextual. For example, someone may tend towards Stage 4 professionally and Stage 3 romantically. Or maybe devolve to Stage 3 based on hunger, fatigue, etc.

Stage 0 is the Incorporative Stage. It’s what infants participate in. Their experience is undifferentiated and they don’t have a “self” to speak of. They’re largely embedded in sensations/reflexes and early attachment. Nebulosity isn’t relevant here because there isn’t really a self to speak of.

Stage 1 is the Impulsive Stage. This typically emerges between ages 2-6. The child is embedded in and identifies with impulses and perceptions. That is, it is the hunger, anger, happiness, etc that it’s feeling. The child lacks the ability to see past their immediate experience, and rules act as an external organizing force to be obeyed or evaded. Nebulosity isn’t relevant here because behavior is largely impulsive.

Stage 2 is the Imperial/Instrumental Stage. This stage often develops between ages 6-15. The child can sufficiently step back from their experience and impulses to coordinate their behavior towards specific goals. The child’s sense of self is organized around needs, interests and concrete outcomes. Therefore, relationships are transactional and other people matter mostly in terms of what they can provide. Nebulosity in this stage is threatening because it challenges the concrete pathways towards win/lose that organize the child’s world. Ambiguous situations lacking an obvious “right answer” that maximizes personal gain feels confusing and illegible.

Stage 3 is the Socialized Mind. This stage often develops post-adolescence. The individual’s sense of self at this stage is deeply shaped by relationships, roles and perceived expectations by their broader social milieu. Individuals at this stage can’t step outside the expectations of their social environment to evaluate them. Nebulosity is tolerable when social consensus exists. However, it rapidly becomes intolerable when consensus of one’s expectations break down, roles become unclear or one’s immediate relationships provide contradictory guidance. This is especially acute when the stakes are high.

Stage 4 is the Self-Authoring Mind. Unlike the Socialized Mind where individuals identify with their social relationships, these individuals identify with their own self-authored value systems. They can distinguish between “what I think” and “what others want me to say, think or do”. This allows them to effectively navigate social pressure by setting positive boundaries. They have greater capacity for navigating nebulosity since their self-authored values provide stable patterns for interpretation. However, this tolerance has limits. Nebulous contexts which genuinely threaten their internal framework, or contexts that are beyond their framework’s ability to solve problems get perceived as personal threats.

Not all adults reach Stage 4. Kegan’s book In Over Our Heads suggests that roughly ~55–60% of adults primarily operate from the socialized order (i.e., not yet self-authoring), about ~1/3 are primarily self-authoring, and self-transforming is rare.

Stage 5 is the Self-Transforming Mind. Unlike the Self-Authoring Mind, the individual can step back from their own value system to see it as one system among many, rather than an absolute truth. They can simultaneously hold multiple systems and perspectives without needing to resolve the contradictions between them. Their own identity and beliefs become objects of reflection and revision rather than objects beyond question. David Chapman describes this as recognizing that systems are “nebulous yet patterned” simultaneously. Individuals at this stage don’t feel the need to reactively collapse nebulosity into premature certainty (i.e. over-control), nor do they feel the need to avoid it altogether via nihilism (i.e. under-control). They can make personal commitments and take action while acknowledging the inherent limitations of any single interpretation.

Realistic example at Google

The description of Kegan’s five stages of development was fairly abstract. Let’s explore how different levels of development may show up in an incident post-mortem at a large tech company like Google.

Premise

• Alice and Bob are both Senior SWEs (L5) on the same infrastructure team managing a caching service.

• Last week, Bob pushed a config change that modified a service’s cache eviction policy and TTL settings.

• This was reviewed and approved by Alice.

• Three days later the service went down due to an unexpected traffic spike from a new client team. Their requests caused the cache to evict too aggressively and it flooded the backend database.

• This caused a 45-minute outage which affected multiple downstream production services and impacted revenue.

• Bob’s change was rolled back and a post-mortem was triggered to discuss the root cause.

Why is this situation nebulous?

Large software projects can be incredibly complex, and the question “What was the root cause?” resists a clean specification. For example, the following competing interpretations about making sense of the incident may emerge during the post-mortem meeting:

• “Bob was careless with the config change.”

• “Alice failed to catch the risky change.”

• “The client team should have announced their load increase.”

• “Capacity planning should have caught this.”

• “Downstream systems should be more robust to such outages.”

• “Our monitoring thresholds are too loose, and should have alerted far sooner.”

Unpacking the patterns of interaction between Alice and Bob

For the sake of demonstration, let’s assume that Bob and the rest of the team are operating at Stage 3 (i.e. Socialized Mind). This follows from Kegan’s research that most adults fail to reach Stage 4. We’ll then vary Alice’s participation between Stages 3 to 5 to see if one person’s developmental stage can shift the dynamics of the entire room, despite Bob being at Stage 3.

Since Bob’s at Socialized Mind, he relies on his social interactions with the group to resolve the nebulosity of “What was the root cause?”. He can’t easily separate “my change contributed to the outage” from “I am a bad engineer and people will see me that way.” Throughout the post-mortem process, he’s very attentive to other’s reactions to interpret how bad this may have been for him. His default approach would be to deflect negative judgement from the group by pointing out that the config change was reviewed and approved. Moreover, perhaps his default would be to over-control by imposing a clear narrative (i.e “not my fault”) onto this nebulosity to protect his social standing, and therefore his sense of self.

Alice at Stage 3

Let’s assume that Alice is also at Stage 3. This implies that like Bob, her sense of self and what she finds valuable is shaped by relationships and perceived expectations. Like Bob, she needs consensus from the group in resolving the nebulosity so that she knows what to think. Therefore, she also hedges heavily, gauge’s the room’s reaction before committing to an interpretation. She feels caught between the tension of maintaining loyalty to Bob (i.e. not throwing him under the bus) versus her loyalty to her overall team (i.e. being honest with what Bob did wrong). Both feel like bad choices without any principled way to choose without group consensus.

Since everyone’s operating at Stage 3, no one’s capable of handling genuine complexity. Bob deflects to protect his reputation and Alice waits for group consensus to organize her actions. Since the group consensus never arrives, the team over-controls to the first narrative that feels socially safe. That is, “Bob will fix it”. Perhaps in the future, a similar incident occurs with a different config change with an eerily similar post-mortem. Bob’s more cautious now and less willing to push for improvements to the overall system. Alice develops resentment that “we never actually fixed the real problem”, and the team habituates an implicit norm of avoiding conflict in ambiguous situations.

Alice at Stage 4

As a Self-Authoring Mind, Alice is guided by an internal system of values. She can clearly distinguish between “what I think” and “what others want me to say”. She can tolerate more nebulosity than Stage 3 because this same internal system stabilizes a pattern of interpretation. But she can’t see beyond her internal system. So her temptation is to confuse seeing a pattern for seeing the pattern, which is exactly what happens. On the other hand, this presents her with a level of clarity that the others in Stage 3 lack.

During the post-mortem she states her analysis directly without worrying about the room’s dynamics. She focuses on systemic improvements rather than pointing the finger at anyone. She doesn’t take Bob’s defensive behavior personally, nor does she feel responsible for managing his emotions. She’s able to coherently defend her case even if the group challenges her interpretation. Her primary tension is between staying true to her internal framework and being aware of the relational cost that acting on “her truth” may entail.

Alice’s clarity allows the team to form a coherent and effective diagnosis, prescription and treatment for the overall system. Perhaps in the future, the system overall is far more robust than it otherwise would have been. But the system remains vulnerable to blind spots in Alice’s narrative. Bob ultimately feels steamrolled and doesn’t push back in meetings anymore even if there’s legitimate concerns.

Alice at Stage 5

As a Self-Transforming Mind, her identity isn’t attached to any specific value system. She can skillfully hold multiple systems/perspectives at once, and doesn’t treat her framework as absolute truth. She’s best able to resist the temptation to over-control and under-control in relation to the nebulosity of the situation.

During the post-mortem she’s genuinely curious about all the perspectives including Bob’s. She asks probing questions to better understand what each perspective was optimizing for as the incident progressed. She’s able to see the incident via multiple frames simultaneously, probes for feedback and is able to see her own view as one among many. For example, “This seems like a testing gap, a monitoring gap and a capacity issue. From where I sit, insufficient testing stands out. But I’ve been burned by that before. What’re all of you seeing?”. The room feels different with her in it. Bob’s unlikely to get defensive because the room’s consensus implicitly shifts from “who is to blame?” to “let’s understand what happened?”. At Stage 3, this allows him to be more honest than he’d otherwise been without feeling threatened.

Alice’s ability to inhabit all of these multiple perspectives without getting too entangled with any of them allows the room to gain more clarity on the situation then they’d otherwise have gotten. She gracefully handles the tension between the necessity of making decisions versus the impossibility of certainty. She can make recommendations for next steps while acknowledging their provisionality.

Perhaps in the future, the team develops a norm of genuine inquiry in post-mortems. Creating safety for Bob allows him to surface warnings earlier next time. This culture of genuine inquiry rapidly improves the process in multiple directions after each incident.

Closing Thoughts

Fault attribution within complex systems is extremely challenging. The inherent stochasticity of AI systems confronts teams with such nebulosity far more often within the process of routine software development. This essay’s scope was to describe the phenomenon of psychological development within adults, and concretely describe why it’s relevant within software engineering. Future essays will ground these ideas more concretely within AI development.

It’s also worth pointing out that adult development isn’t purely individual. Organizations create environments that shape the developmental trajectories of everyone inside them. They establish norms, expectations, and patterns of interaction that either support or hinder people’s capacity to navigate nebulosity. Conversely, the behavior of people within organizational containers shapes the overall norms within the organization.

Some organizations habituate wisdom within their participants. Others engender increased foolishness. For example, how a company/org/team runs their post-mortems gradually cultivates or erodes the developmental capacity of everyone in the room.

The next essay explores what it means for an organization itself to orient itself towards development.

Acknowledgements

Brian Whetten, Prof John Vervaeke, Charlie Awbury, David Chapman for everything they’ve taught me.

Dan Hunt for helpful discussions, editing and generally co-creating these essays with me.

Stochastic systems resist the clean specifications and deterministic fixes that knowledge workers are trained to expect. This discomfort shows up not just at the individual level but in the collective psychology of teams and companies. They tend to react in one of two ways when it hits.

Discomfort at the level of teams

Over-control: Believing you can eliminate all “bad” model behavior

Suppose an embarrassing LLM output gets escalated within a team. Instead of calmly accepting that this is what stochastic systems do and thinking carefully about how the underlying evals and quality control processes should get updated, there’s a frenzied overcorrection. People start saying things like “This must never happen again”. What follows is a predictable rush to inflate the evals with examples that capture the relevant edge cases. Or to add “hacks” in the LLM’s scaffold to attempt to deterministically catch such behavior. None of this is done carefully because it’s done in a high-stress state. But the team manages to put out the fire for now.

The stochastic system keeps producing embarrassing outputs, albeit each time with a different flavor. The evals are patched haphazardly to account for these embarrassing outputs without much concern for the overall actionable questions that the evals were originally designed to answer. The entire system gradually becomes a mess that’s difficult for the team to effectively iterate on. This creates even more stress for the team.

There’s often some trade-off in any stochastic product in terms of explicitly minimizing embarrassment (i.e. failures) and maximizing value (i.e. victories). One extreme way to avoid a specific type of embarrassment is for the system to never try at all. For example, by punting on many user requests with the response “I can’t help you with this request.” Over time, despite what the team says, the evals increasingly skew towards avoiding embarrassment rather than amplifying customer value. It’s difficult in such cases to point the finger at any single individual. The overall system gradually converges to this equilibrium due to the contributions of every single participant.

Despite the entire system tending towards this extremely risk-averse attractor, the team is constantly asked to create more value for its customers. This only adds to the collective stress. So the cycle continues.

Under-control: Offloading value judgments to the model

The increased nebulosity of AI products forces teams to grapple with what “good” means far earlier in the product lifecycle. Even if the product is entirely owned by a given team, defining and defending what “good” means is hard. It requires working with competing values both within the team and between its external collaborators. Reaching strong inter-subjective agreement may only be possible after substantial inter-subjective disagreement. It may require uncomfortable conversations where specific individuals have to take formal accountability for what “good” means. So people avoid doing it.

LLMs are the perfect accelerant for these avoidant tendencies. For example, it’s common for people to set up LLM-as-a-judge systems to evaluate LLM workflows that they care about, without creating corresponding evals for these judges. This looks rigorous because it’s produced by a “machine” or “algorithm”. But it just sweeps the problem under the rug. The judge needs its own evals but no one recognizes this. One could ascribe this to ignorance. But often, it’s because people are reluctant to invite and metabolize disagreement.

The result is that the team is gradually disempowered to truly take ownership for its product’s quality. They incrementally create a system they don’t understand and can’t control. Nevertheless, they’re constantly asked by their superiors to create more value for customers. It becomes harder and harder to slow down and course correct as the team grows disempowered. They’re in a perpetual state of overwhelm. So the cycle repeats itself.

What this looks like at the company level

Over-control: Preserving familiar structures

Change management is complex, difficult and frightening. Many organizations correctly identify that injecting more nebulosity into their system carries scary risks. But fear of these risks can cause executives to unskilfully grip control too tightly. That is, to create increasingly compelling rationalizations for why necessary change is actually unnecessary. Or perhaps, to pay lip service to the necessity of creating change without actually doing much.

There may be substantial excitement about AI in leadership meetings. But leadership keeps funding the same projects as before. Only this time AI is bolted on top. Strategic planning treats AI as a tool to accelerate existing roadmaps rather than something that may fundamentally change what’s possible or what should be built. The entire leadership team is already perpetually swamped. So their revealed preference is that they don’t have any time to look into this new AI thing. But they also don’t want to give off the impression that they don’t know what they’re doing. Or that they’re not taking this change seriously.

When someone proposes a genuine restructuring around AI capabilities, they get labelled as naive or reckless. Safe incremental improvements get praised instead. These rationalizations sound reasonable precisely because they contain a kernel of truth. Consequently, the company moves incrementally faster along the same workflows that it had before. Model improvements fail to unlock novel possibilities for what the company can accomplish in the world.

But the world isn’t static. The landscape starts shifting as the models get better and competitors start changing their operations. At this point, an intervention that seemed too radical doesn’t seem radical enough. But it goes against the deeply held belief to play it safe. This tension creates a lot of stress, and the cycle continues.

Under-control: Performing competence instead of developing it

It’s scary for leaders to show vulnerability. AI is possibly one of the largest shifts since the Industrial Age. So it’s especially scary for leaders to admit they don’t know how AI works or how their business can create value from it. It’s also scary for many teams to feel that their leaders don’t have all the answers.

Nevertheless, everyone can see the shifting patterns of the world. There’s a feeling that something needs to be done. It’s common for such leaders to deputize subordinates that they can’t properly evaluate, precisely because they don’t know much about AI. This also means that they lack the discernment to distinguish genuine competence from charisma and bullshit. This inevitably amplifies the performative dynamics within the company and within their AI projects specifically.

Mundane AI workflows get rebranded as “agents” to match the broader external hype cycle. Sacred cow AI projects get created. They’re quickly branded as “critical”. Evals and performance metrics for these various AI projects eventually morph to manage leadership’s emotions more than tracking value created for customers. Various cultural structures emerge to reinforce these dynamics. For example, vacuous Slack channels are created to “share AI wins” that serve as opportunities for ambitious people to “manage up” rather than truly disseminating valuable tips and tricks within the company.

Moreover, the “critical” AI projects quickly become boondoggles because they lack authentic strategic direction. Their overall dynamic starts to attract ambitious people seeking visibility rather than people best suited to do the work. The whole thing gradually becomes a political shitshow. Inevitably, things often go wrong with such projects. When it does, there’s a profound absence of leadership to course correct.

Feedback from customers doesn’t reach the right levels of leadership in sufficiently undistorted form. The organization can’t learn because various critical corrective feedback loops start to become too broken. The same ignorance that creates all of these performative dynamics also prevents the organization from recognizing and correcting it. The longer this theater carries on, the worse this feedback loop becomes. So the cycle repeats itself.

What I think is going on

AI places greater demands on an individual’s capacity to relate to nebulosity. A similar dynamic also takes place with the collective psychology of teams and companies. Each of the patterns above is what happens when teams and companies lack sufficient capacity:

• Over-control takes place when teams and companies reactively seek to collapse this inherent nebulosity into something that feels more controllable. They attempt to relate to stochastic software with processes and operations as if it were deterministic software. That simply doesn’t work. Each “failure” to hold down control creates more and more stress until something “breaks”.

• Under-control takes place when teams and companies avoid relating to this nebulosity altogether. They attempt to create as much distance as they can between themselves and this new reality. But this doesn’t work either and creates its own distortions. Similarly, each of these “failures” simply raises the temperature of the system creating more stress until something “breaks”.

The transition from deterministic-first to AI-first software is profound. Creating value in tandem with improvements in model performance requires the cultivation of new capacities. These demands to sit with greater levels of nebulosity get expressed at the level of an individual, team and company. It turns out that there’s already some research from Robert Kegan and others that attempts to articulate how such capacities could be developed. The next few posts will explore this body of work to better understand the “map”, before we dive in with “solutions”.

Acknowledgements

Brian Whetten, Prof John Vervaeke, Charlie Awbury, David Chapman for everything they’ve taught me.

Dan Hunt for helpful discussions, editing and generally co-creating these essays with me.

Despite the uncertainty, most people do feel the profundity of the societal change we’re about to experience, even if they can’t consciously articulate it. There’s an undercurrent of anxiety about who’ll be affected next. At the same time, frontier models are still hard to use in most use cases and not yet ready to wholesale eat into labor budgets. They make mistakes a competent human wouldn’t make.

It’s natural for people to be wondering: when will the threshold get crossed? Two years? Five years? Has it already happened in some domains and we just haven’t heard about it yet?

This is a confusing and uncomfortable situation to be in. It’s not clear what it means to remain economically valuable as a knowledge worker. The skills that got you here likely won’t automatically carry you forward. I’ve worked through my own version of this discomfort and come out the other side with some clarity.

Four failure modes for AI knowledge work

I’ve noticed the following consistent patterns in how people are relating to AI.

Not engaging at all

The org-wide Slack channel about AI tools sits unread. You skip the optional training. When someone asks what you think about the latest model release, you deflect. “I haven’t really looked into it yet.”

It’s not that you’re opposed to AI per se. You’re just too overwhelmed to engage because you don’t know where to start. But there’s also a quiet fear underneath. You’re worried that something specific about you means that you’re going to get left behind by this transition. So if you try to use AI you’ll look like an idiot. Especially when there’s so many confident people on Twitter or at the office. It seems easier to just wait until things settle down. Perhaps until there’s some stability of “what it all means” and someone can tell you exactly what you need to learn.

But this creates a gap between you and the people that are actively experimenting. It starts to gradually widen. You freak out every time you notice this gap, which makes the stakes for experimentation feel even higher. The end result is that you continue waiting, and the cycle continues.

Doomscrolling the AI news cycle

You can recite the differences between GPT-5 and Claude 4. You know which models are best at coding vs. writing. You diligently track all the latest benchmarks and follow all the right people on Twitter. You sign up for all the latest demos the moment they show up in your feed. It feels like you’re staying current but it’s deeply performative.

There’s a gap between knowing about AI and knowing how to use it. Your consumption doesn’t have much depth. While you may have opinions about each model, you’ve got no muscle memory for effectively working with them. Doomscrolling social media creates a sense of motion. But it doesn’t change much in your day-to-day ability to create value.

You choke and panic when someone finally asks you to build something with AI. You realize that you don’t actually know how to wield this technology to create value. All that forward motion confused urgency with reactivity. This realization is especially painful because you have been putting in the effort. Just not in the direction that develops the skills you need. Confronting this is really overwhelming. So you go back to doomscrolling and the cycle repeats itself.

Surrendering to the model

You paste in the task, hit enter, copy the output. You ship it if it looks good enough. The model seems so capable and confident that critically engaging with its outputs feels like a waste of time. Who are you to second-guess it?

Moreover, being picky about what “good” means is hard because it requires judgement you’re not sure you have. It’s easier to just let the model decide.

But this surrender is extremely disempowering. You gradually become interchangeable with anyone else with the same tools. You’re not actually using nor developing your idiosyncratic judgement, which is the very thing that would gradually make you asymmetrically valuable. In the meantime, the models continue to improve really fast. You feel so far behind the eight ball that spending any effort cultivating discernment feels like a waste of time. So the cycle repeats itself.

Jumping straight to building an AI-first product

You quit your job to build an AI startup. Or you volunteer to architect the company’s new sacred cow AI project. You want to prove your worth as a leader and an expert. That you “get it”. And you’re okay to “fuck around and find out”. You figure you’ll just throw yourself into the deep end and figure it out as you go along. The future seems glorious with possibilities.

Six months later, the product doesn’t work reliably. You’re hit with the realization that you don’t actually understand why the models behave the way that they do. For example, you got really swept up by the grandiose vision of vibe coding breathlessly described on Twitter. Or the capabilities of “agentic search”. Your vision was too ungrounded in reality. You jumped way past the kiddie pool despite having no prior experience with any comparable technology. In the meantime, you substantially overpromised what your project would be able to deliver. Perhaps your investors or the rest of the company already made difficult-to-reverse decisions based on your confidence. But again, the product doesn’t really work reliably and you can’t see a clear path for fixing this. Coming into contact with the realization that your visions may not come to pass, along with the pressures of expectation produces an incredible amount of stress on your body that creates feedback loops of reactivity.

Building an AI-first product is extremely risky and capital intensive. It’s still very poorly understood and very difficult even for strong teams. Most teams will likely fly too close to the sun. Making good decisions in this domain requires practice and a substantial amount of leadership.

What I think is going on

All of these responses are deeply understandable. I’ve participated in some versions of all of them and reflected on why they occur.

Knowledge work is shifting from “how” to “why/what.” Deterministic-native work placed an extensive premium on being a “machine” that executes the “how” within the assemblage of a company’s workflows. For example, writing code, fixing bugs, shipping features. Moreover, even though there was nebulosity about “why” and “what” a given project should have been doing, the deterministic nature of software meant that one could confidently make forward progress via iteration.

AI-first work is fundamentally different. It forces teams to confront the nebulosity of the “why” and “what” in a comparatively confronting way. The inherent stochastic nature of LLMs necessarily front-loads a lot of consideration about what “good” even means. At the very least, so that this “good” can be codified into a prompt. This process immediately triggers questions like:

• What should this agent do when a user asks something ambiguous?

• What counts as “helpful” in this context?

• What’s the right trade-off between safety and usefulness? Is there a way to simultaneously optimize for both of them?

Deterministic-native software afforded a sense of certainty in the “how”, even if the “why/what” were deeply uncertain. The stochastic nature of AI-first work requires embracing a lot more nebulosity than what orgs have been built to handle. This creates a specific kind of pressure. Individuals have to accept that while certainty is impossible, you can develop the skills and discernment to confidently navigate product development faster. Not because you eventually achieve certainty. But because you get better at navigating without it.

Each of the patterns I described above is a way avoiding this new reality of increased nebulosity:

• Not engaging avoids the discomfort entirely. But the world moves on. You fall behind without noticing until the gap feels insurmountable.

• Doomscrolling creates the feeling of keeping up without ever having to sit in the ambiguity directly. You know about AI but never face the discomfort of making it work.

• Surrendering avoids having to define “good” by letting the model decide. But then you’re not adding value. You’ve outsourced the very thing that would make you irreplaceable.

• Overreaching tries to skip past the uncomfortable learning phase. But you can’t. The discomfort isn’t a phase. It’s the territory itself.

This is why I’ve come to see the pursuit of value creation with AI as the cultivation of capacity for nebulosity. Such capacity is a lot easier to cultivate if your broader culture supports it. The next essay explores what it looks like for the collective psychology of teams and companies that face their own version of this challenge.

Acknowledgements

Brian Whetten, Prof John Vervaeke, Charlie Awbury, David Chapman for everything they’ve taught me.

Dan Hunt for helpful discussions, editing and generally co-creating these essays with me.

There’s a deeper pattern underneath these frictions. Our community is confused about what it means for something to be an “agent”. When everything is an “agent”, nothing is. This and subsequent posts will explore a very specific sort of agent. One that maintains state across time, continuously learns from its environment and presents itself almost as a “coworker” in social environments. This is in contrast with other notions of “agent” that are either rigid workflows implemented with LLMs or enhanced information retrieval tools. Admittedly, we didn’t have models good or cheap enough for this confusion to matter until now. But we’ve started to get there with Gemini 3, Opus 4.5, GPT-5.1-Thinking, etc.

Before exploring the implications of such agents, we need to define them. That’s what this essay does. Subsequent posts will dig into the harder questions.

Amusingly, as is often the case, my response to this confusion is to provide my own definition. An AI-first agent is an AI-first product that also satisfies the following criteria.

1. Each agent instance has a clear “boundary” that separates what’s “inside” the agent and “outside”.

2. Each agent instance has the capacity to skilfully request extensions to its boundaries based on its ongoing internal processes, so as to create more value in its context. Conversely, reductions to its boundaries causally lead to an immediate shift in its ongoing internal processes, so as to continue creating the most value that it can.

3. Each agent instance can participate in social dynamics to change its behavior for the pursuit of value creation. For example, feedback from an agent instance’s “manager” might lead to immediate changes to the deepest levels of its 4P stack, like its name. But feedback from a fellow employee or another agent may not.

4. Each agent instance has an approximation of the 4P stack in its memory that allows it to better navigate its environment. Please refer to Professor John Vervaeke’s work for more context on this. This post contains a brief explanation of his work. More specifically:

   1. Propositional knowledge - it has a collection of propositions that may not be in any of the underlying LLM’s weights, but that nevertheless condition them to better fit within the contexts they find themselves in.

   2. Procedural knowledge - it has a collection of “skills” that it can deploy based on what it finds relevant within a given context.

   3. Perspectival knowledge - it maintains a repertoire of “episodes” that inform what perspectives in its environment it should find relevant.

   4. Participatory knowledge - it provides the agent with a coherent identity, overarching hierarchy of values, and a description of the agent-arena relationship that it’s participating in (that is, how the agent understands itself in relation to its environment and the other agents within it).

Notice that these criteria address the frustrations I opened with. The inability to communicate over Slack or gather requirements stems from the absence of persistent boundaries and social participation. The missing context about tacit social dynamics reflects an impoverished 4P stack.

My previous post shows that Claude Code isn’t yet an AI-first product. Therefore, it’s necessarily not an AI-first agent either. But it’s worth imagining what a version of CC might look like if it satisfied some of the criteria specified above. The most conspicuous gap is the absence of a stable identity. Without one, the agent cannot have clear boundaries because there is no persistent self to bound. It cannot participate in social dynamics because it lacks the continuity required for others to hold it accountable over time. And perhaps most importantly, it cannot develop participatory knowledge. A coherent identity, a stable hierarchy of values, and an understanding of one’s relationship to the arena one is participating in all require something that persists across interactions. Today’s CC has some memory via features like CLAUDE.md. But by default, it doesn’t continuously learn from each interaction. It cannot refine its sense of self over time.

Such an identity would start with something like a Google/Microsoft account: a name, email address, and a trusted description of its human owner. Possessing such a stable identity would also make it easier to deterministically enforce various rights and sanctions upon the agent. A stable identity would also make it easier to potentially give it access to a payment method for participating in the broader economy, and the ability to provision other identities (e.g. GitHub account, GCP account, etc).

Stable identities would make it easier to afford clear boundaries of “inside” and “outside” for the agent. For example, all of its private Drive files, emails, GitHub repositories, etc would be “inside” the agent’s boundary. It’d then use existing social environments like email, Slack, etc. to communicate with other human or AI agents that are “outside” of it. The agent’s governors and human collaborators would be able to reuse security semantics from each of these products. Ditto for its AI collaborators. One could also imagine a central dashboard for the agent’s owners to examine its current tasks, resources, thoughts and goals.

Developing software with such an agent within a broader social milieu would increasingly resemble developing with a human coworker. But one that possesses a lossy copy of all human knowledge. Its owner would give it specific permissions to the GitHub repo and shared Drive, and then ask it to go off and create value within the company.

Different “roles” of humans and agents would have different levels of social privileges to request/demand changes to its observable behavior via natural language instructions. For example, the agent’s owner would have the right to change its name, change its tasks, etc. But perhaps a human collaborator would only have the right to “request” it to join a specific piece of work. This would afford its human collaborators the ability to build cultural technologies to better govern its behavior/alignment.

I’ve described what might seem like a distant future. But each of these capabilities sits upon a spectrum, and we’re closer to viable versions than you might think. One can imagine substantially stronger and weaker versions. It’s not clear exactly when such a system will hit the market. But I’d anticipate some rudimentary MVP certainly within a year or two. I myself see concrete paths to bringing such an MVP to market in the next year with modest capital. Therefore, I know that many others do so too.

The arrival of AI-first agents will have profound implications for the world even if they’re not “AGI”. The next few posts in this series will map out some key open questions that arise. Most of these questions defy obvious answers or solutions. I invite you on a journey to sit in these questions with me.

• What is the relationship between such AI-first agents and a company’s culture?

  • How can such AI-first agents be used to afford the cultural development of organizations?

• How can we invite the social sciences into AI? What would it mean to develop a sociology of AI employees interacting with humans?

• What would the governance of such agents look like?

• Rather than thinking about an “end state”, what are some key thresholds that we should seek to monitor? Coding agents are a leading indicator of what’s happening. What’s happening to coding agents will soon take place for the rest of knowledge work.

• What would it mean for an agent to develop genuine procedural and perspectival knowledge through experience? Today’s agents have pre-loaded skills. But they don’t acquire new skills through practice. They don’t accumulate episodes that inform what they find salient in novel situations. What changes when they can?

We’re extraordinarily early in AI-first development. This post demonstrates how minor tweaks can make most AI products like CC substantially more AI-first. Specifically, this post introduces a “reflective memory” that allows CC to reflect and update its memory based on past interactions with the user.

Claude Code obviously satisfies the first criteria because the LLM Claude entirely drives it.

Starting in version 2.0.21, CC began to somewhat satisfy the second criteria when it started presenting multi-choice questions for unclear requests. Crucially, this process of requesting clarifications falls short from our aspirations because it doesn’t seem to learn from the user’s responses.

CC somewhat satisfies the third criteria because it affords a rudimentary notion of “memory” via CLAUDE.md files. It’s good that they’re vanilla markdown files. This allows CC to easily update them. It’s also good that the /init command somewhat attempts to build a representation in CLAUDE.md of the user’s overall social milieu (i.e. the codebase). However, /init isn’t very sophisticated and lacks many product integrations. CC also doesn’t proactively update CLAUDE.md by reflecting on its own mistakes and misunderstandings in relation to the user’s instrumental goals.

CC’s elegantly simple design makes it straightforward to extend in this direction. CC stores partial logs as JSONL files in ~/.claude/projects for each working directory. These logs can be mined via bulk inference to analyze misunderstandings between Claude and the user, which can subsequently be used to update CLAUDE.md.

This DoubleAscent/claude-logs-analysis repo provides a sample implementation. A collection of Python scripts read these JSONL files and perform bulk inference via a MapReduce pattern.

1. It runs bulk inference via a “mapper” that runs an analysis prompt over every single JSONL file. The mapper’s context window contains both a sanitized JSONL file and the current CLAUDE.md.

2. These inference results are then concatenated and fed into a “reducer” prompt. This produces a final report about what aspects of CLAUDE.md should be updated.

3. The user then manually prompts CC to read this report and subsequently update CLAUDE.md.

We’re essentially setting up a feedback loop to run “prompt optimization” on our revealed preferences as we use CC, with CLAUDE.md’s contents acting as the “learnable parameters” of this process. We’d likely run this process everyday or every week. We’d expect the contents of CLAUDE.md to rapidly reflect changes in our preferences and behaviors. This makes CC overall far better at tracking misunderstandings (criteria 2) and integrating these misunderstandings into its memory (criteria 3).

Note that this repo was intended as a demonstration of what’s possible. One can imagine many extensions/improvements to make the feedback loop spin faster:

• Create a slash command to make all this more seamless.

• Create a UI to make it easier to quickly arbitrate the various misunderstandings and provide context for each one.

• Make the MapReduce robust to larger inputs via sharding.

• Use a better model for analysis than Gemini Flash Lite.

• Create evals for the mapper/reducer prompts and hill-climb on them.

• Hook up the results of AskUserQuestion to a centralized memory.

• Have a shared CLAUDE.md for a team, and learn a consolidated set of norms across the whole team.

If linear chains of causation characterized the old world, then feedback loops and spirals will characterize this new AI-first world.

Please leave a comment or email me at varun [at] doubleascent.com if you try out this workflow or anything like it. I’d love to hear from you!

Subscribe now

A trained human admin wouldn’t have made these mistakes. They would’ve intuited what was important, had enough context about the named entities to transcribe them correctly, and would’ve recognized how to make the notes most legible to me.

I’d have felt seen by a trained human in ways that existing note-takers can’t emulate. More broadly, most AI products I use in my day-to-day workflow can’t see nor learn about me over time. They see a thin archetype of me, not me. They either churn out slop or shove me straight into an uncanny valley.

It’s tempting to conclude that the latest models are incapable of such collaboration. But today’s frontier models are more than capable of powering incredible experiences. Product builders lack the right conceptual frameworks for thinking about AI-first products.

Errors in deterministic software usually show up as bugs you can fix. In contrast, errors in stochastic software (i.e. LLMs) show up as misunderstandings you can only resolve iteratively. An AI system often fails at its task because it lacks relevant context and makes a bunch of implicit assumptions that turn out to be wrong. There’s no deterministic method of a-priori resolving such misunderstandings. It’s a necessarily iterative process. It’s impossible for both users and product builders to write complete specifications ahead of time that anticipate all such misunderstandings. For example, even the user might not be able to enumerate every situation ahead of time in which the voice recorder should highlight or deprioritize a sentence. Taking this seriously implies a profound shift in how AI-first products should be designed. Rather than treating the model as a static component that either works or doesn’t, the entire product is framed as an evolving conversation of testing value between the system and its users.

This insight can be formalized into three criteria for whether or to what extent a product is AI-first.

1. The product’s core user journey relies on LLM inference calls.

2. The product is in an ongoing process of surfacing and resolving misunderstandings between itself and the user. It’s mindful of the tolerance window of this specific user for this specific interaction, given the type of workflow that the user is engaged in.

3. The product maintains and uses a rich, long-term memory of the user and their social milieu in service of value creation.

The voice recorder’s frictions are easy to place when re-examined through the lens of these criteria. The recorder clears the first criteria because it already uses AI in its core loop. But it totally fumbles the second and third. It doesn’t have a good way to surface or resolve misunderstandings about what the meeting was really about, and almost no memory or orientation toward my actual world, so I end up with slop instead of something that feels like notes from a Chief of Staff.

These criteria point to a very different worldview and corresponding ontology for products that were simply impossible pre-LLMs. In this worldview, the system stops acting like a static tool that executes a formal specification (i.e. computer program). Instead, it behaves like a dynamic process that formulates, tests and falsifies hypotheses about what’s most valuable to its users based on evidence.

Imagine a voice recorder that, after three months, knows your standups need ticket numbers but your 1:1s need emotional subtext. That notices when a conversation shifts from routine to existential and adjusts its summary accordingly. That learns your threshold for clarifying questions and stays just under it. That tracks not just what you said in meetings but how your priorities have shifted across them, and uses that to anticipate what you’ll find valuable before you ask. In that world, the product is increasingly made of feedback loops rather than static features. It uses the flexibility and rapid improvement of frontier models to keep probing for opportunities in value creation.

Zooming out shows that the same pattern applies to other AI products. A sales tool that is AI-first keeps updating its sense of which relationships and deals are truly consequential. A coding assistant that is AI-first keeps learning your codebase, your style and your team. As more people use these systems, they can generalize in two directions simultaneously. That is, they build personal models of individual users and population-level models of value, then test which patterns actually transfer. They increasingly become scientists in the discipline of economic value creation with integrity.

Satisfying all three of these product criteria is profound. It yields a system that feels “alive” and “responsive” because it’s actively testing for what a user finds valuable. The most valuable AI-first product for an individual is one that keeps trying to fit itself to that user’s evolving conception of value over the longest possible time horizon. The same holds at the population level. The most valuable product is one that quests for value for an entire population over the largest possible horizon while still seeing each individual in front of it.

We’ve barely begun to scratch the surface of what all of this means. AI-first products sit on a spectrum based on the degree to which they satisfy the three criteria. All three criteria are interlocking and interpenetrating with each other. Rapid model improvements will yield possibilities that are difficult to imagine right now.

Will you build products capable of absorbing these improvements? Or will you keep bolting LLMs onto product surfaces that feel alienating?

Please email varun [at] doubleascent.com if you’re actively building AI-first products. I’d love to hear and learn from you!