New Year’s is a terrible time to stick to a resolution (or follow a new strategy.) Spring is not a great time to be cleaning. The air is warmer, the days are longer. I don’t want to be cooped up in an attic, going through the remnants of old hobbies. I want to be sitting on patios drinking margaritas with the people that I love. It’s much more effective to spend the time between years cleaning and leverage spring energy to propel change in your life.

After Christmas, there is a natural rhythm for cleaning up and reorganizing your physical space. Rolling up Christmas lights, stuffing them back in the closet. figuring out where to put new gifts you received, and what to get rid of to make space for them. It’s the last week of the year, the closest we get to the capitalism meat grinder taking a beat.

This period is perfect for clearing the deck and closing loops. Wrap up those projects. Get ready for tax season. Getting the house in order physically, mentally, and spiritually. In Japan, they call it Oosouji, “big cleaning.”

Then, January rolls in. It’s dark. It’s cold. The elements test you with the shittiest weather, rain 3-5 degrees above freezing. All of the chill of snow without any of the fun. I would choose a tornado over a 36℉ storm. I force myself up at 5 am to make it to the gym. It is beyond crowded. I do what I can. I’d work out outside, but you know, the rain.

I stay home. I open the work laptop. A 12 encircled red greets me. An inbox full of people who had the audacity to circle back in the new year.

I press on. February rolls around. Statistically, the worst month for seasonal depression. Whatever seasonal joy I had in the reserves is gone. If I was doing good on eating habits, too bad, combo breaker: Girl Scout cookies are back. I am not trying to be my best self right now. I am trying to survive.

And I always do. The sun rises, soup season comes to a close. People crawl out of their blankets. I was walking to my car from the doctor’s office, and a woman at the crosswalk, apropos of nothing, looked at me and says ‘wow it’s so beautiful out today!’ There is an energy in the air.

When the sun rises, so does serotonin. A 2017 study in Current Biology by Stothard et al. showed that humans’ circadian rhythms shift earlier in spring. People wake up more energized, more biased towards action, with more daylight to act on it.

Trying to be resolute in January is working against nature. Doing so in spring is working in harmony with it.

And as a bonus, the gym will be less crowded.

In April, I wrote about some of the techniques for putting my Leeroy Jenkins ethos into action. The reception was positive, so here are five more. These are centered around building momentum and getting started when you're stuck.

1 - Break Inertia

The hardest part is getting started, so make getting started dead simple. If you are stuck in a doom loop, move. Physically move. Stand up. Flap your arms. Anything to get you physically moving. Not starting a project? Do anything. Sketch a note. Archive an email. Create an empty file. Anything! It’s all about pattern breaking.

2 - Just Push Buttons

Don’t know how to get started? Start. Action produces information. You will learn more by doing than by theorizing. I’m reminded of a quote from Ricki Heicklen about their quantitative trading school:

So when people walk through the door, the very first thing we do is have them start trading. Absolute number one, immediately just throw people in without explaining the rules, without explaining too much logic about what is all the syntax going to look like? We say, “Hey, we’re going to go around, play this trading game,” and we get them started and correct as they go. And that’s because a very very big part of our pedagogical philosophy is that the best way to learn how to trade is to trade.

“You need sufficient knowledge to get started” is a self-limiting belief. Getting started is how you know what to do next.

3 - Ask an AI

If you have no idea where to get started, an LLM can get the juices flowing. But to the first point, be wary of planning disguised as procrastination.

I wouldn’t take any plan at face value, but if you don’t know what to do next, brain dump whatever facts, wants, desires, and constraints you have, then prompt your model of choice(if you don’t have a model of choice it’a Opus 4.6) with something like:

I would like to (achieve this goal, build this product, write an essay about this). Study (my notes) Interview me and ask me questions to build out (a step-by-step implementation plan, list of potential actions). (that plan should include A, B, C)

Tweak as needed, then go through the exercise. Don’t give the AI solutions. Give it problems, and see where it goes. Even if the plan is bad, the questions could get your brain moving and spark other ideas. You’re exploring, not handing agency over to a flanker.

4 - Moments over Systems

As James Clear says, “You don’t rise to the level of your goals, you fall to the level of your systems.” While there’s truth to this, it can nerd snipe people into believing “you need a system to get started.” You don’t. System or not, you can make decisions and seize good moments.

Systems help with long-term change. But they hurt when you dodge short-term wins because you’re fixated on some bigger fix you’ll put into place “one day.”

You can pass on that glass of wine with dinner. You can order the grilled entree instead of the fried one. You can make one good decision in a single moment. You don’t need a diet plan, a nutritionist, or an app with macros and charts. You can be good without being perfect.

And besides, what is a system but a way to make the good choices more consistently?

5 - Spontaneous Deep Work Sprints

Ive read many productivity and management books, and one of the most consistent pieces of advice I see is this: the secret to getting a lot done is long, uninterrupted time blocks. They suggest you book appointments with yourself and put them on your calendar. Start the blocks like you’re going to the movies: use the bathroom and grab a snack before you lock in.

I’ve tried this several times, and it never sticks. Some productivity tricks work for some people, but personally, I have never been able to take “an appointment with myself” seriously. Hard as I try, I never stick to the work block. The other challenge is distractions: I’m on the manager schedule, and things come up needing immediate attention. Other times it’s slow, where I’m monitoring and on-call, ready to hop in where needed.

What works for me is grabbing time blocks when I can get them. I don’t schedule them, but when I have a deep work task, and I spot open time on the horizon, I decide then and there to lock in for 30, 45, or 60 minutes. The shorter timeframes are a catalyst: How much can I knock out in a sprint? I don’t think knowledge workers sprint enough, going as hard as they can for a short burst.

“Sprint” has lost its meaning in tech. You cannot sprint for 2 weeks. A sprint should feel fast, feel hard, feel uncomfortable. You should take a break after a sprint. Consider a working HIIT session.

It’s ‘moments over systems’ in practice. I can’t reliably block off time, but I can be opportunistic about the time presented to me.

Conclusion

Putting these together, I realize struggling to get started is being trapped by the Nirvana fallacy: You don’t start today because there could be better conditions tomorrow. When in fact, these are lies we tell ourselves, mental labor we bear. If we were to look at it, we would see that it is pointless. We can throw it away. We can walk lighter, freer.

You don’t need motivation to start; you need to start to build motivation. You don’t need a system or a plan to take action; you need to take action before you have a system. You don’t need to make an appointment with your future self; you just need to move.

It’s that time of year. The final week, where the worst versions of ourselves plan for the better versions we’ll be next year. We crawl out of the pile of discarded wrapping paper and plates. The background drone of holiday films begins to fade. The time to lock in approaches.

We all know resolutions are a fantasy. Most fail to lead to permanent change. Resolutions feel like a burden, like an exam you’re going to fail instead of a book that shows you a new way of thinking. You hit the gym for a week or three, then start to miss. It becomes a stressor. You toss the resolution in a desk drawer and ignore it until Q4 reflection time.

Last year, I found a different approach: I didn’t go in with a New Year’s resolution. I went into 2025 with a New Year’s strategy.

A resolution is a statement of a goal. Strategy is a framework that shapes decisions. Resolutions are rigid, strategies are flexible. Resolutions are narrow. Strategy is holistic. Resolutions lack a system for dealing with changing contexts. A strategy helps you decide what you’re going to do, how to prioritize, and empowers you to make decisions with confidence.

How to write a strategy

The word ‘Strategy’ conjures images of McKinsey consultants billing $600/hr to create slide decks. Decks with lots of charts. Charts that will do more harm than good. Strategy need not be highfalutin, and unlike McKinsey consultants, can deliver real value and change. Here’s a dead-simple summary of what I mean by strategy, riffing on the ‘kernel’ idea from ‘Good Strategy / Bad Strategy.’ A strategy has four parts:

1. Goal - What do you want to accomplish?

2. Diagnosis - What’s wrong?

3. Guiding policy - Your approach, what to do, what not to do.

4. Coherent Actions - How can you put this into practice?

That’s it. It doesn’t need to be a huge document or grandiose idea. It could be a paragraph. It could be a sentence. WWJD is a complete strategy.

(Tech bros made that weird, too, by the way)

An example

Imagine you're driving from San Francisco to Seattle, and you want to get there in under 12 hours. That's your goal. For those unfamiliar with the West Coast geography, 12 hours is a theoretically doable timeframe but the variance can be high. It's possible to lose 2 hours if you pass Portland at the wrong time, and there are several other traffic bottlenecks.

A possible strategy to achieve this somewhat optimistic goal would be to use multiple drivers to enable continuous driving while optimizing around traffic patterns rather than absolute distance or speed.

The strategy isn't "take I-5" or "drive fast", it's a higher-level approach that introduces new possibilities (can drive straight through), adds constraints (need to coordinate schedules), and impacts other decisions that might be made (no need for a sleeping stop). It's a set of fundamental approaches which generate a cascade of tactical decisions: when to swap drivers, timing of departures to avoid city traffic, length of rest periods, etc.

Dan Pupius, You need to be more strategic - A primer on strategy for software engineers

At the start of the year, I declared 2025 my 'year of actionable thinking', and I crafted the 'Leeroy Jenkins Strategy':

1. Goal - Spend less time overthinking, more time on execution

2. Diagnosis - Too much time spent on planning and infotainment.

3. General policies - Be intentional about planning and research. All information should be actionable. Favor taking small next actions and focus on progress over perfection.

4. Coherent Actions - See Jenkins Strategems, Just Push Buttons

At the beginning of the year, I was also moving from IC to engineering manager. I didn’t know what the job would entail. But having a strategy meant I could apply my original goal to novel areas. I could learn what questions to ask in meetings to drive decisions and keep us out of a planning bog.

An anti-example

My stated goal of publishing 10 posts in December. I had a goal, but lacked a diagnosis. I hadn’t considered the amount of holiday planning and activities, the ‘get it in by Q4’ work push, the general crud, the persistent cough all parents of young children have in winter. The sun, despite all the negative feedback, continues to set at 4:47pm. How can I be expected to hustle when the sun can barely put in a full day’s work?

3 ways to keep yourself honest

Once you have a strategy, you have to ensure you put it into action. Wisdom is realizing this video is sincere:

What helped me:

Regular check-ins - If you take away one thing from this post, let it be this: If you want to focus on a goal for the long term, set up a system to bring your attention to it regularly. Add a recurring task to your to-do list. Slap a sticky note on your bathroom mirror. Wear a charm or bracelet.

Strategy-driven preparation - What can you do today to enact your strategy in the future? Strategy can help you in the moment, but also beforehand, in planning and preparation. Map your route before you leave.

The Focusing Question - I’m a stan for Gary Keller’s The One Thing. Though I’ll admit you can get 90% of the value by asking yourself the “Focusing Question”, both regularly and when you don’t know what to do next: “What’s the one thing I can do such that by doing it everything else will be easier or unnecessary?”

What do you want to accomplish next year? How could you write a strategy to achieve it?

Maybe I should make a WWSLaBD bracelet 🤔

Writing increases your luck surface area

“When you write more, you create more random opportunities” - Zizek

Writing on the internet is an incredible economic engine for individuals. It’s a kind of force multiplier and serendipity engine. Blogging is a way to build your network, reach more people with your ideas, create connections and also stand out from the crowd. It’s an unfair advantage with an activation cost of $0. — Tom Critchlow, The Three-Body problem of consulting

The internet runs on words. Writing is required for effective work. — Ash Ambridge

Writing is a social practice

1

“Build the things that will allow you to engage an audience and improve it over time. Included in this: Friendcatchers. Friendcatchers are small, contained lists, essays or apps that solve resonant (emotional, relevant), tractable, underserved problems.” — Swyx, Friendcatchers

Shoving writing in people’s faces is cringe. Sharing things you’re actively thinking about, working on, interested in, and passionate about is a great way to connect with interesting people. - Tom Critchlow

Writing can generate interesting conversations, create new connections, and strengthen existing friendships.

Writing can be an act of service. Write things that help others. Write for an audience of one, then send them something via DM.

Writing is a long-term game

This is one of my beliefs: there has never been a more vital hinge-y time to write, it’s just that the threats are upfront and the payoff delayed, and so short-sighted or risk-averse people are increasingly opting-out and going dark.

If you write, you should think about what you are writing, and ask yourself, “is this useful for an LLM to learn?” and “if I knew for sure that a LLM could write or do this thing in 4 years, would I still be doing it now?” – Gwern

When looking for jobs, or thinking about a new product, or anything else, you always wish you had written more. This isn’t something you can do just-in-time; you need to do it before it’s time. It’s a game with short-term risk and long-term payoff.

Writing pieces are assets. They can and have led to landing clients, getting speaking gigs, and getting jobs.

Intrinsic benefits of writing

writing makes you see there are holes in things. I’m never sure what I think until I see what I write. And so I believe that, even though you’re an optimist, the analysis part of you kicks in when you sit down to construct a story or a paragraph or a sentence. You think, ‘Oh, that can’t be right.’ And you have to go back, and you have to rethink it all.” - Carol Loomis

A careful and honest writer does not need to worry about style. As you become proficient in the use of language, your style will emerge, because you yourself will emerge, and when this happens you will find it increasingly easy to breakthrough the barriers that separate you from other minds, other hearts - which is, of course, the purpose of writing, as well as its principal reward. - Shrunk, Elements of Style

Expressing our own thoughts in writing makes us realise if we really thought them through — Sönke Ahrens, How to Take Smart Notes

Creative writing is self-care for ideas. — Ash Ambirge

Writing codifies your ideas, then you can communicate them more clearly. Everyone else is out there talking in first drafts.

Writing helps you develop a voice, a tone, a style.

Writing is a form of learning. The best way to learn something is to explain it to others.

Writing is an infinite craft. You can stop improving your writing in the same way a pro golfer can hit an 18. In a fast-moving world of technology, writing is an evergreen practice. I will write until I’m no longer able.

Writing is an exercise in getting thought plaque out of your head. It’s good for your mental health.

On writing prolifically

If a post comes out about the right thing at the right time, it will go viral. If it does not, it will not. All you can do is keep releasing — Venkatesh Rao, The Calculus of Grit

Writers have a terrible sense of which posts will be popular with readers. […] Because that’s hard to predict, and because the impact of a single hit is so extreme that it’s worthwhile to up the odds, and doing that means writing more. — Unknown source from a dead link

Further Reading

• Good X thread: If you have a technical blog, and it’s brought you some positives (job-related, social, etc.), and you don’t mind your tweet being screenshotted, I’d love to hear about them.

• Tom Critchlow - Small-b blogging

• Henrik Karlsson - A blog post is a very long and complex search query to find fascinating people and make them route interesting stuff to your inbox

I sometimes can't tell when AI makes me more productive or slows me down. When I'm learning or brain rotting. So I started collecting case studies, doing research. Then I found this article: How to use AI without becoming stupid. Eureka moment. Now I have a clear line in the sand for deciding whether an LLM is up to the task.

Vaughn Tan Rule2:
Do NOT outsource your subjective value judgments to an AI, unless you have good reason to—in which case make sure the reason is explicitly stated.

This matches my experience & that of my coworkers. So what does it mean in practice?

Use LLMs to inform, not decide

Don't ask "What should my next career move be?" Instead, ask "Give me examples of career strategies for people in my scenario." LLMs gather information, humans make the call.

One big source of alpha: CLI-based tools like Claude Code and Codex as plain text code search engines. Customer support asks, "Do we have a setting to disable the transaction email with {subject line}?" I work for a company whose backend is written in Haskell, a unique language with a steep learning curve. A curve I've never bothered climbing. Now I have a much easier time understanding how things work.

Use LLMs to edit, not interpret

LLMs excel at data transformation. LLMs are great at turning text from format A to B. Haskell to English. Code to customer support. I've used it to generate plain English summaries of JSON objects. You can take a screenshot of anything and turn it into text or a CSV.

Now that I'm in a more managerial role and writing on the side, I find this kind of 'text scripting' incredibly useful. These are cases where judgment calls aren't needed. When it is, I find a way to stay on top of the process.

I don't trust LLMs to write; they lack voice3. Instead, when writing this piece, I will have an agent copy a draft, then make edits based on a style guide. Afterwards, I diff both pieces and put together a final version. Fewer mistakes, more humanity.

However, I would be wary of using it for summarization. Some text processing is sneakily a value judgement: You are deciding what to include and what to omit. What to emphasize and what to downplay. You are also making decisions about structures and prioritization. These are the kinds of writing work I wouldn't outsource to an LLM.

LLMs don’t always write good code, but code doesn’t always have to be good

Tell me if this resonates: You see the hype around AI delivering developer productivity gains. You see the takes on social media calling it bullshit. You try it at work. The result? A tool that ignores instructions, reimplements existing code, churns out slop. You spend more time writing prompts and reviewing than coding. You've hit the paradox of automation.

The faster code can be written, the more crucial human judgment becomes. Real savings come from code that requires less judgment.

Day job code requires working with a large codebase and trying to achieve a specific result. It requires subjective judgment on which parts of the codebase are important, and what the output should look like.

Sometimes code doesn't matter much. Prototypes. Command line utilities. Utility scripts cost nothing now. CLI-based LLM tools write and run them. Contrast coding at work with "vibe coding," an app, where you care less if the final outcome is a little different than what you were expecting. It's all jazz, baby.

LLMs drastically changes the math on automation.

I think I could whip up any single-purpose script in 30 minutes or less. Automation costs roughly a pizza.

The biggest gains aren't in optimizing existing workflows. They're expanding what's possible. With a tool that writes & executes scripts, what wasn't worth doing before?

Product engineers and designers can now open PRs to production. Should companies encourage this? Why or why not?

We live in a world of infinite content. There are more books, movies, shows, songs, games, essays, and works of art than you'll ever see in one lifetime. And on top of that, we have never-ending “for you” feeds vying for our attention.

On the bright side, infinite quality exists, too. You have to look for it, pay attention to it, curate it. It takes effort. You have to maintain a healthy information diet1. Less junk, more nutrients.

You get out what you put in

From Taste is the new intelligence: "what you consume, what you engage with, what you amplify—becomes a reflection of how you think."

Curation is a part of self-expression. Part of being yourself is curating your taste, which requires being intentional in what you read & watch, and reflecting on what resonates with you and why. What you consume shapes how you think. Output is a function of input.

Less junk

Remove as many ads as possible.

It's worth upgrading streaming services. Judiciously unsubscribe from email spam. Set up ad blockers or a Pi-Hole. You choose what you consume, not corporations. Gentle reminder that advertising works on everyone, especially people who think it doesn't work on them.

Keep your follows small and high quality.

You should avoid social media, but when using it, keep the number of accounts you follow small and high-quality. Don't let noise fill up your feed. You choose what you consume, not influencers. They are called influencers for a reason.

Have an outcome-focused approach to content

If you want to be entertained, watch entertainment. If you want to learn, study. Don't half-assedly listen to a podcast while you're doing dishes and tell yourself you're "learning." The whole genre of "infotainment" exists for the smart person who likes the feeling of learning, regardless if they retain any useful information. No judgment, liking that kind of stuff, I do, but I'm honest with myself that it's for entertainment purposes only.

More nutrients

Prefer Lindy content

If you had to guess, what percent of content you consumed today is less than one year old? Less than one month? Prefer content that has a long shelf life and took a long time to make. Books and essays over tweets and vertical videos.

Prefer uncommon content

Forage for cult-classics and hidden gems. More obscure sources lead to more original thinking. Hideo Kojima got ideas for the tone of Death Stranding from an album he found in an Icelandic record store. He hired the band to work on the soundtrack. Breakout videogame of 2025, Clair Obscur: Expedition 33, is heavily inspired by a French sci-fi novel that didn't sell enough copies to warrant an English translation.

There is nothing wrong with consuming mainstream content; it's popular for a reason. But you'll be less inspired if that's all you ever see.

Recommendations over algorithms

Ask friends for recommendations, or read books that your favorite authors reference. My reading list for the last decade has been:

1. I see a recommendation for an interesting article or book

2. If I like it, I dig deeper into articles and books recommended in the piece.

3. GOTO 1

Don't let a few algorithms decide what you're going to consume. And be part of the solution! Share more about what you love! Support small-time creators! The world needs more recommendations, fewer algorithms.

Take time to process

Take time to reflect and integrate what you've learned, instead of jumping from one thing to the next. Instead of starting the next chapter of an audiobook on an evening walk, give yourself the space to think. It's overwhelming out there. We get exhausted if we don't give ourselves space to process all that we've seen.

AI and skill issues

Elden Ring introduced many players to the Soulslike genre, a notoriously difficult type of video game. When newcomers complain about unfair bosses and punishing mechanics, they may be told to 'git gud.' or simply 'skill issue.' - gamer slang meaning "the problem isn't the game, it's you."

The first boss lampshades this, reminding the player after defeat they aren't yet ready for what lies ahead. New players, myself included, heard this phrase over and over and over again.

At least, they aren't ready yet.

Some people take offense at being told "skill issue." But truth is, its empowering. You can't change the game, only l how you play.

Same with LLMs.  People say they don’t get the same results others rave about, and think its baseless hype.[^1] Unlike Elden Ring, it's harder to tell what is skill and what isn't. Elden Ring is difficult, predictable, and fair. LLMs are a non-deterministic blackbox. The same prompt can give different answers, and it can be unclear why. I still don't know how to tell the difference between user error and model strangeness.

Skill issue on my part I suppose.

Defining AI fluency

If some people are better at using LLMs, what skills set them apart? What does it mean to be “good at AI”?

This internal AI fluency framework from Zapier has been making the rounds. I don’t know if I agree with all of it, but it’s a decent starting point.

Part of fluency is knowing what’s available, what tools can and can’t do, and matching the right model to the right task. Skilled operators write stronger prompts, and they also pick the best model for the problem. That’s the difference between someone typing two sentences into a free ChatGPT window, and someone who studies the different elements of what goes into a prompt. See intimidating chart:

But ultimately, I think the deeper skill is being able to figure it out. By the time the charts are made they’re already outdated. Instead of diving into byzatine documents, you have to have to just push buttons, evaluate the results, and adjust your approach.

Many AI adoption strategies fail, but not for the reason you think

I’ve seen this headline thrown around a lot: "MIT report: 95% of generative AI pilots at companies are failing"

Some people read that as “don’t run a GenAI pilot.”(AI is hype!) Others see it as “figure out how to be in the 5%.”(I gotta git gud!)

Pro tip: When people cite a "studies show X" headline, go find the study and read it your self. In today's infinite content environment, this is a superpower.

Go ahead, give it a shot: The GenAI Divide: State of AI in Business in 2025, tell me your biggest takeaway after reading it.

Leave a comment

My take: it’s more "enterprise companies struggle to adapt to new technologies" and less "AI fails to deliver."

Losing companies are:

• Overspend on sales & marketing, underinvest in back of house automation.

• Buying expensive, tailored, inflexible GPT over generics like ChatGPT and Cursor

• Try to roll their own internal AI systems.

• Have complex enterprise processes and refuse to change them.

Winning companies are:

• Solving for learning, memory, and workflow adaptation.

• Build systems that learn from feedback and retain context/

• Invent new, high-leverage processes that leverage AI's strengths.

What stuck with me: the big wins aren’t coming from making current workflows faster, but instead creating entirely new ones.

A better question than "how can we go faster?" is "What can we do with LLMs that wasn't possible before? "

I recently found the *Complex Systems* podcast, a where Patrick MacKenzie talks investing, AI, building software. An oddly high number of guests are former professional poker players. It could not be more by shit. In a July 2025 episode, In July he talked to former Jane Street employee Ricki Heicklen. Ricki founded Trading Camp, an immersive four-day bootcamp where people learn quantitative trading through simulated stock games, team projects, and lectures.

The Line:

Ricki: So when people walk through the door, the very first thing we do is have them start trading. Absolute number one, immediately just throw people in without explaining the rules, without explaining too much logic about what is all the syntax going to look like? We say, "Hey, we're going to go around, play this trading game," and we get them started and correct as they go. And that's because a very very big part of our pedagogical philosophy is that the best way to learn how to trade is to trade.

The best way to trade is to trade.

Later that weekend I was thinking about tinkering with Claude Code—and regretting how much time I’d spent talking about it, watching YouTube, and reading articles without learning the tool. The exact kind of behavior I’ve been trying to curtail all year long. I took Ricki’s advice. I installed Claude Code, bought $20 in API credits, and started pushing some buttons.

Diving headfirst into Claude Code

At first it felt daunting. What project should I build? Which model? What belongs in my local CLAUDE.md file? How will I verify correctness? How do I prompt it? I’d stall there. I even keep a note with twenty “LLM Project Ideas,” yet when I sat down, I didn’t know where to start. After the episode, I figured, “start anywhere!”

I had a few changes I wanted to my personal homepage. I use Claude Code and it was able to get the job done! I wanted to try something bigger. Then, I stumbled across this post: “The 7-Day Roguelike That Took Six Minutes.” Perfect! I love roguelikes, and this is a decent sized greenfield project.

It took hours, not minutes. When I got stuck, I read docs and searched for answers. I kept a running list of open questions and filled them in as I went. I wasn’t “studying for someday” anymore—the mythical day when I’ll “have time.” As a parent, the only thing more fantastical than a dungeon of skeletons and orcs is the idea that time will appear out of nowhere. I was looking for specific information to remove specific roadblocks, so I could get back to work.

The next week at work, I shipped more pull requests than I had in the previous month—thanks to tricks, tactics, and prompts I picked up while experimenting. The best way to vibe code is to vibe code. You just gotta push buttons.

Pushing buttons in fast-moving fields

Someone asked me for AI course recommendations. I said I’m skeptical of books or courses. because the field is moving fast, and LLMs are an always morphing black box. What worked last week may not work this week. By the time anything that long gets published, it’s dated. You have to get comfortable taking the path less traveled, ordering off-menu, going off-book.

There’s plenty of alpha out there if you’re willing to just push buttons.

But systematic note-taking isn’t the problem: It’s our emotional relationship with information. Negative patterns often lead to self-defeating behavior.

Here are some of the anti-patterns I’ve observed:

Note taking without intent.

Your work and goals should shape what you consume and collect. Information is only useful when it’s actionable.

If I didn’t write, I wouldn’t keep an Obsidian vault. Once I asked what my notes were for, it changed what I saved and what I ignored. I started keeping more data, quotes, and personal stories—the kind of raw material that strengthens an essay.

Deciding what to keep is an exercise in taste.

It clicked when I created a separate notebook just for work. Having a notebook with a singular focus made it clear what belonged and what didn’t. Now opening that vault helps me switch into manager mode, and I don’t get distracted by unrelated information.

I reach for notes after outlining an article. I open my work vault every day and use meetings & project notes to set my intention for the day. I know what notes to save because I have a clear sense of when I’ll use it and how.

Digital hoarding

Collecting notes without intent is just digital hoarding. A quote from ‘I Deleted My Second Brain’:

For years, I had been building what technologists and lifehackers call a “second brain.” The premise: capture everything, forget nothing. Store your thinking in a networked archive so vast and recursive it can answer questions before you know to ask them. It promises clarity. Control. Mental leverage.

But over time, my second brain became a mausoleum. A dusty collection of old selves, old interests, old compulsions, piled on top of each other like geological strata. Instead of accelerating my thinking, it began to replace it. Instead of aiding memory, it froze my curiosity into static categories.

You don’t need to capture everything. Some ideas are better left forgotten.

Good notekeeping includes pruning. There is a point of diminishing returns. When a note is no longer actionable, let it go2.

Lost in the system sauce

Note-taking systems are a great opportunity to do some deck chair moves and pretend you’re being productive.

I’ve tried PARA3, I’ve tried no folders, and I’ve tried dozens of folders. It waxes and wanes, and ultimately, it doesn’t matter. As long as the notes are helpful to you, that’s all that matters4.

Modern note-taking apps are cursed by their inspiration. Tools like Roam and Obsidian were inspired by Sönke Ahrens’ book, “How to Take Smart Notes,” and his “Zettelkasten” — a note-taking system that revolves around paper cards and a numeric indexing system connecting them. This system easily translates to hyperlinks, but it misses the point. We have search bars.

Don’t get trapped by taxonomies. You don’t need a complex set of tags, folders, data-views, and two-way wikilinks. All you need to do is include keywords so you can find what you need when you need it.

Tricking yourself into thinking note-taking is writing.

I’ll admit, I’m guilty of this one. This is the exact kind of behavior I am trying to rid myself of when I set my New Year’s resolution. Notetaking is a means, not an end. It’s easy to get lost collecting, organizing, and refining notes—without ever shipping anything.

Work isn’t useful until it helps someone else. Writing in your notebook does not count.

Treating notes like work instead of play

Note-taking becomes more fun when you treat it like play. Don’t take it too seriously. Mess around. Give notes weird titles. Be like the kid in school who used seven different marker colors. Doodle, draw, diagram. Use your notebook to think, not just to record.

Notes as weight, not fuel

Ever opened your notes and felt dread? You keep telling yourself you’ll organize them one day. But every new entry adds to the pile.

Now imagine this: you’ve just spent a couple of hours cleaning your house. It’s immaculate. You feel proud and relieved. You feel lighter, calmer. Notes can feel that way too.

When I’m feeling down and burnt out, I have found that cleaning and organizing a physical space makes me feel better. Similarly, when I’m feeling stuck, going through notes, rearranging them, removing clutter, and finding new patterns can help me get unstuck. Your digital gardening should leave you feeling energized, not drained.

When I’m feeling burnt out, organizing physical space helps. Notes work the same when I’m feeling stuck. Rearranging them, tossing old ones, spotting patterns—it’s like clearing a mental fog. Digital gardening should energize you, not drain you.

Replacing your primary brain

One quote from 'I deleted my second brain’ stuck with me:

"A quote would spark an insight, I’d clip it, tag it, link it - and move on. But the insight was never lived. It was stored"

Writing things down can short-circuit the experience of living them. But this cuts both ways: The ‘Zeigarnik effect’, happens when holding thoughts in our short-term memory causes increased stress. Whenever I feel like I have a lot of tasks on my plate, sketching out a quick to-do list helps. Leverage your notes to clear your head for deeper thinking, then use that space.

On the other hand, we can use notes to bring back the feeling we had before, and reignite inspiration. I’ll leave you with a quote from Kendrick Lamar, about how he sees his notebook:

Take it home, Kendrick

"I have to make notes because a lot of my inspiration comes from meeting people or going outside the country, or going to the corner of my old neighborhood and talking to a five-year-old little boy. And I have to write them down and then five or three months later, I have to find that same emotion that I felt when I was inspired by it, so I have to dig deep to see what triggered the idea.. It comes back because I have key little words that make me realize the exact emotion which drew the inspiration."
- Kendrick Lamar

You saved time coding, sure, but you spent more time speccing, reviewing, and testing. You aren’t sure you saved time at all. This new method makes you feel a little uneasy in a way you can’t quite put your finger on. You’re still not sure what the hype is about.

I’ve been there. That’s when I realized I was seeing AI coding at it’s worst. If you’re only using AI code on large codebases, you’re missing out. AI coding shines on small projects. Little Keurig Cups of code. Utility scripts. Micro apps. Tools with no business on a production server. Disposable software.

Writing disposable software

ChatGPT sometimes gives the perfect example of disposable software: you give it a task, it writes Python code, executes it, gives you the answer. You can cajole it in this direction by explicitly saying “use python code” to solve this. ChatGPT intends for code that is only ran once. But it’s cheap and fast enough to product that it’s okay.

AI Coding shifts the answer to the “is it worth the time?” test. More often now, the answer is “yes.”

I wanted to build out a tool for writing my product thinking book: at this stage of editing I am constantly adding, removing, and rearranging entire sections. I wanted to keep each in a markdown file, and then assemble an ebook based on a table of contents file. Claude code was able to whip this up for me in about 15 minutes. Otherwise, I’d have slogged through a verrrry long Google Doc.

I could see a lot of my writing projects looking like this in the future: repositories that are a mix of markdown files and Python scripts that help with formatting, data viz, etc.

Disposable software sidestep’s AI codings biggest flaw: the code is often ass. Keurig coffee will never beat a hand-pulled espresso. But here, quality is irrelevant, only results matter. It’s like the hotel room coffee on a business trip, the one you make in your Keurig. The on you drink to muster the energy to venture into sunlight in search of a decent espresso.

Disposability is a feature, not a bug

Trying to make disposable software production ready is a trap. Someone tried to make Vibe Coding Enterprise Edition and it did not work. This is why I like this method for prototyping. Another trap I saw back in my coding days was someone taking a weekend prototype, and then trying to evolve it into a real app, instead of starting from scratch.

Putting it into practice

As you go about your day, take note of rote tasks you could automate. Jot down small product ideas. When writing prompts, consider your options. You could ask ChatGPT “hey solve this for me” or you could say “write a python script that solves this for me.” Maybe it’s even something temporary, like an application to help you manage your Christmas shopping.

Then depending on the scale, you can later try to one-shot them, or go through a full vibe coding cycle of spec → write → review until you have something workable. It’s cheap to try. If it fails, you can always throw it away.

(This piece is part 8 of the Prompt Protocols blogchain)

Like social media, AI is a new digital infrastructure layer. It's going to create new opportunities, and destroy us in many eerily similar ways.

AI will build pave new roads to riches. Social media became a launchpad for influencers and small business, making it easier to reach more people. AI is doing something similar to product code, content, and products. Creation, like reach, is faster, cheaper, and more accessible than any point in human history. Influencers are already pivoting to renting out their voices to train models.

Infrastructure isn’t free

Social media offered reach—but at the price of rising ad fees. Coding assistants like Cursor are making coding faster, but also more expensive. They are currently charging $50/mo, but those prices won't last. I’ve been thinking a lot about this quote from Geoff Huntley’s recent article:

What happens when AI prices indie devs out? The gap between what a major corporations and small teams can produce will only grow.

And there’s the creeping dependency, almost an addiction. I’m shifting from anxiety about AI writing code for me… to realizing I might not be able to work without it.

Wake up babe! New psychosis just dropped!

AI will mess with our brains in new and inventive ways. It’s a well-studied phenomenon that constant social media connection increases depression, anxiety, and loneliness in teenagers. Optimized-for-obsession algorithms bewitch us into scroll trances, draining our time and attention.

Users who use character-based chatbots also show an increase in depression, anxiety, and also increased delusion. Our caveman brains aren’t ready for voice bots that are indistinguishable from humans.

New terms are emerging—“prompt compulsion,” “hallucination anxiety.”One guy made an AI employee for his startup and then wanted to fuck it. Wonder what we're going to name that one?

Announcing DictatorGPT o4-mini-high coming this winter

AI is a dream tool for authoritarianism. he more we rely on chatbot summaries instead of exploring the web ourselves, the more centralized information becomes, and the easier we are to manipulate and control.

Social media was already powerful weapon. Algorithms took tiny impulses and magnified them into extremism. Social media made men lonelier, then build a lonely male → Manosphere → alt-right radicalism pipeline for them.

Pop-quiz hot shot!

Here’s a quick 1-question test to determine where you stand in this pipeline:

• Level 0: You have never seen this video

• Level 1: You have seen this video

• Level 2: You know other people have been politically radicalized by this video

• Level 3: You are politically radicalized by this video

Authoritarian regimes rely on centralized media. Imagine if a chat bot was your primary source of truth. A recent hack of Grok led the bot to constantly rant about South African conspiracies. What happens when people start doing this more subtly, and on purpose?

Slop is the Fascismcore aesthetic

AI was baked into Trump’s 2024 campaign aesthetic. his Truth Social account posted AI-generated images of Taylor Swift fans and that bizarre “Trump Gaza” video.

Any tool that empowers creators will be co-opted by extremists. We’ve seen the decline of the United States as a global super power and it looks like slop.

So what’s next?

And don’t worry—soon we’ll have AI friends on Meta, so we can enjoy the worst of both worlds.

The question of AI’s impact isn’t if, and it’s not when anymore. It’s: what the hell is this thing going to do to us?

Most decisions are satisficing decisions

I recently had to decide which projects my team would tackle over the next two months. We had several good options, but how could we know which was the right bet? We had informed guesses, but design needed enough runway—so we had to choose quickly.

Pretty much any option would have worked. We could have thrown darts and been okay. We listed our best guesses on impact and feasibility, and picked our next three projects.

Most questions don’t have a correct answer, but all questions have a best answer — subjective and based on context. The best choice is simply the best one you can make in the moment.

Selection sort all the things

I learned this from a beta copy of a book my friend Swizec is working on. Humans are terrible at comparing a lot of options, but we’re good at comparing two at a time. When you’re unsure what to pick or prioritize, make a list and selection sort it.

(God these videos are so satisfying)

Draw blurry pictures

There are two JPEGs compression algorithms: baseline and progressive. Here’s how they render:

Baseline JPEG

Progressive JPEG

Be a progressive JPEG. When you're working on something, start with a minimal but complete version. Better to have all the pieces, regardless of how flawed they are, than an unfinished piece.

There are several benefits to this approach:

1. It informs the work —the middle oftens informs the end, and the end informs the beginning.

2. You’ll likely hit “good enough” sooner than expected.

3. Complete rough drafts are easier to share and get feedback on.

4. It gets you to the revision phase faster. Quality comes from the edits. The trick to delivering high-quality work is a willingness to do more rounds of revisions than most people.

What’s a 10-minute thing you can do right now?

I love this list of high-agency questions from Chana Messinger. The whole thread is gold, but I want to highlight this one in particular:

There’s usually a smaller step you can take to move things forward or a simpler version of what you want to do. When stuck writing an article, I’ll ask myself “can I write a 250-word version of this idea?” That’s enough to get started.

This pairs well with the blurry picture approach. My new favorite way to finish a piece is to set a series of timers1.

• Can I make in outline in 10-minutes?

• Can I flesh it out and order it in another 10?

• Can I get together a complete shitty first draft in 30 minutes?

In under an hour, I’ve gone from an idea to something I can work with2.

Finally, have some fun with it

Thanks to work travel and other internal writing projects, I haven’t posted on here in nearly a month. I felt overwhelmed trying to get back into it, especially with the growing list of articles I feel I “should” finish. Shouldn’t I focus on the most important, biggest ideas first?

I carved a section of this Substack out specifically for informal, off-the-cuff writing. But I felt I “shouldn’t” write one of those, because I had “more important” pieces I should work on instead.

The problem is, that makes this feel like work. I already have work from 9-to-5 and it’s plenty. I start avoiding this text editor entirely. The whole attitude is silly if I think about it. No one is forcing me to write here, or meet any kind of editorial guidelines or deadlines except the ones I set for myself.

Writing is generative, and sometimes fun is what gets you unstuck.

Sometimes the best way to be productive is to spend 20–30 minutes on something frivolous. It reminds me why I enjoy this in the first place. The bigger stuff can wait. they’re easier to write after I gave myself a little time to do a small stakes writing sesh.

Maybe it’s productive procrastination. But — it builds momentum. This resolution isn’t about optimizing productivity or getting more done.

It’s about getting and staying unstuck.

I used to hold back on hobby projects because I thought it would ruin the fun. Turns out focusing in, trying to do my best, and putting effort into learning and deliberate practice, make the hobbies richer and more enjoyable.

Burnout isn’t caused by overextertion. It’s caused by lack of meaning, boredom, or a lack of control. Building skills helps you push through those. More effort equals less burnout.

At the start of the year, one of the goals I set for myself was to put more effort into writing, to see how that felt. While I am still early in this journey, I am already seeing the benefits. The biggest one? Writing feels less draining. The cost-per-article is lower now — I can move from idea to article faster and with more confidence.

My dad gave me a guitar for my 13th birthday. I didn’t ask for one, never wanted one, but he knew I would love it. At first, it wasn’t fun. My ring finger on my fretting hand is a centimeter off, so instead of hearing a satisfying crunching power chord I hear a dissapointing thunk. I play the same simple songs over and over and over. How many times do I have to slog through Bush’s Glycerine I can play it all the way through?

Then, it clicks. I gain fluency. I can just play. Chords and melodies come natrually. I can nail the opening riff of Black Dog and Master of Puppets without thinking. I can jam and groove with friends. It’s amazing. Dad was right.

All it took was three years of consistent effort.

Passion is built, not discovered

Effort and deliberate practice increases your enjoyment and connection with an activity. As Magness and Stulberg point out in their book The Passion Paradox, being connected and intentional doesn’t drain you, it fuels you.

I don’t know why it took me so long to realize this. Besides guitarm I’ve put effort into other hobbies: I’ve run a half marathon. I’ve defeated Malenia. I enjoyed them, despite it being hard. Maybe it’s because after a stressful day, it’s harder to pick these hobbies up. We do only have so much energy in a day, after all.

Life is a marathon, not a sprint. Hell, as I approach my 39th birthday, some days are a marathon. We have to leave something in the tank so we can bring the best of ourselves to what matters. It’s an investment, and it’s worth it.

That said, as soon as I finish this draft manuscript of System is as System DoesI’m taking a break and not looking at the damn thing for a week.

What would you say you do here?

I’m not a boss, I’m a talent wrangler.

On film sets, talent wranglers manage schedules, handle logistics, and keep actors focused while coordinating between production crews and studios. Their job is to get the best performance out of my actors given the role they are playing in the movie they’re making.

Similarly, my job is to get engineers to perform at their best. I manage schedules, handle projects, maximize reports’ focus time, while coordinating between product, design, and stakeholders.

What does that look like day-to-day?

Many people misunderstand managers' roles, assuming we don't contribute much. When asked, managers throwing out vagaries like “unblock” and “liason” doesn’t exactly help.

To help answer this, I started keeping an inventory of the tasks I was doing from day-to-day:

• Project management: - Creating tickets and project plans, and assigning tasks. Running kickoffs, retros, and syncs.

• Switchboarding: - communicating across multiple slack channels and DMs, maintaining and index of resources so I can the right information to the right person at the right time.

• Hiring: - writing job descriptions, conducting interviews, and coordinating with recruiting and finance to get resources.

• 1-on-1s: With my team, with my manager, and EPD2 peers, with my manager.

• Performance management: Developing and enacting growth plans. identifying opportunities that align with team members’ goals. Asking other people “What would you say they do around here?”

• Writing: PRDs3, SOPs4, agenda, TPS reports5, miniature manager manifestos. Especially working remote, written communication is a huge part of the job.

To summarize, my job is to do all the context-switching so that engineers don’t have to.

But don’t you miss writing code?

This is the second most frequent question. “Building is so fun! Why would you trade that for [shudders] meetings?”

There is still some “IC work” involved, at least for now. When I do I focus on work that unblocks and empowers my team: Reviewing tech specs and pull requests, taking care of the occasional urgent bug so my engineers aren’t interrupted.

Also, I get to do a more abstract, more fascinating kind of engineering. I am still working on a system. The system is made up of values, processes, goals, and plans. We are constantly designing, refining, and refactoring it. Instead of writing code for computers to execute, I write ideas for people to implement.

Gambler: Secrets from a Life at Risk is the autobiography of Billy Walters. I was lucky enough to grab this on Libby when there were no holds on it. The book chronicles his journey from being born in the sticks of Kentucky, spending a few decades as a shithead, getting his shit together, and becoming “The Micheal Jordan of sports betting.”

I enjoyed listening to Billy’s stories about growing up in Kentucky and gambling in Vegas. He reminded me of sitting on the porch on a hot Georgia summer, eating watermelon with my Grandpa, and listening to him spin yarns about his past. Full or flourish and exaggerations. How much is true? It doesn’t matter. I know Billy didn't really put “50 stacks of newspapers” in his bike basket for his paper route at 9 years old. But it makes for a heck of a story.

Why people gamble more when they can’t “afford” it

Walters started from meager beginnings and spent his early years in a cycle of hustle and self-destruction. Have the biggest quarter of any car salesmen in the region, then blow it all on booze and playing pool.

Keeping money in the bank when you aren’t bringing much in is tough. You could spend March, April, and May scraping together a few extra bucks, only for your in-window AC unit to die in June. Your savings are gone; you’re sweating from the sun and your next car payment. And it’s not just about the money. It’s about the habits.

It’s rational for people in that situation to put a few bucks towards lottery tickets, not just for daydream equity, but because a 1-in-11,688,053.52 chance of escaping poverty is better than zero.

There’s a strong connection between growing up in poverty and having a gambler’s spirit. “gamb00l,” as we called it back in the online poker boom of the early 00s. You see people with steady incomes getting nowhere. When you can’t afford to fund a ROTH IRA and your employer does not provide a 401k, saving is not seen as a path to wealth.

The lucky few find opportunities to channel this gamb00l into variant-income careers like sales, finance, and entrepreneurship. Maybe they get started waiting tables. The difference between success and failure is whether or not you can break old habits and build new ones.

Betting like a businessman

When Billy watches one of their friends die in his 40s, it’s a wake-up call. He stops smoking, drinking, and going to casinos(though he continues to bet on sports). He starts saving his money.

Eventually, he partners with a group of quants called ‘The Computer Group’ and manages to build a sustainable, winning sports betting operation. Walters claims to bring in $50 to $60 million in winnings in a good year.

When he starts “wagering as a business,” he uses mathematical modeling to create lines and bet the deltas between his lines and the sportsbooks. Model shows the Bengals as a +4.5 favorite and the sportsbook has them at +2? Put some money down and pray Joe Burrow pulls through for you.

Funnily enough, this book made a lightbulb go off with me about investing. Value investors do this with stock pricing—betting on your models vs. the market.

Investing lessons from a gambler

1: Bankroll management — Have a bankroll(a pile of money dedicated solely to your betting). You have to be comfortable losing 100% of this money. Don’t put more than 3% on a single bet. And when you’re ahead, take some money off the table.

2: Every percentage point matters — The house edge on most casino games is 3-5%, but that’s enough for casinos to rake in $8.3 billion last year. The slimmest edges can make the difference between winning and losing.

I’m reminded of a quote from Roger Federer’s commencement address at Dartmouth:

In tennis, perfection is impossible... In the 1,526 singles matches I played in my career, I won almost 80% of those matches... Now, I have a question for all of you... what percentage of the POINTS do you think I won in those matches?

Only 54%.

In other words, even top-ranked tennis players win barely more than half of the points they play.

3: Find the best price — Pay attention to fees and taxes. Use multiple accounts, and when you’re ready to make a play, ensure you’re paying as little as possible.

4: Different horses for different courses — There is no universal strategy. Understand the context you’re in, and adjust accordingly.

The Gambler’s journey

I want to say Billy had an unconventional path out of poverty, but that feels weird to say because there isn’t a conventional one. Some standard advice still applies: If you’re not born lucky, you must work hard, have discipline, and still get lucky at some point.

I’m reminded of the time in college I spent two years making a living playing online poker: People imagine gambler as these rogue mavericks who put it all on the line. In reality, the successful ones are stoic, risk-averse, nerdy number crunchers.

And they had a couple of lucky breaks early on, which seeded their bankroll and their self-confidence.

1 - James Sinegal, Founder of Costco

Sinegal runs one of the most successful retail operations in the company. Does so selling products nearly at cost. Pushes suppliers to make packaging with less waste. Has a company minimum wage of $19.50. One threatened to kill a man over hot dog pricing.

2- Mike Illitch, founder of Little Caesars Pizza

He ran a local youth hockey league. Founded the Little Ceasar's Love Kitchen, a food-truck-based charity that provides free hot n' fresh pizza to the homeless and people in disaster areas. Paid Rosa Park's rent for a decade and didn't tell anyone. He knows he could charge you more than $5 for a pizza, but he chooses not to ensure low-income families always have access to a hot meal.

3 - “Trader” Joe Coulombe

Paid will above market wages. Invented almond butter. Includes a treasure hunt game in every store to keep your kids entertained. Makes it so you can grocery shop and still not have to cook.

The common thread

Besides lacking the dead eyes you see in many other CEOs, what do their business have in common?

They all pay their employees above-average wages. They all have a keystone value item. Because of these three men I can have a dinner of a hot dog, coke, pizza, and wine for $8.50. Hockey stick growth for my cholesterol for under ten bucks.

But what’s most interesting is that each of them competed on price without sacrificing quality. Costco does it on volume and a no-frills shopping experience. Little Caesar’s does it by keeping the menu lean. Joe achieves this by sourcing non-traditional products and packaging them under his brand.

Value for your customers. Generosity for your workers. Focusing on the essential.

That’s the kind of place I’d want to run.

People’s assumptions of how writing works are completely backward. You don’t have an idea, make the klackity noise, and ship it. Writing and sharing are what create ideas. It’s not the end of a creative process. It’s just the start. I call this generative writing. I’m amazed at how much can come from writing when you put more intention and elbow grease into it.

The not-so-subtle art of giving a f*ck

1

Earlier this year, I committed to putting more effort into increasing both the quantity and quality of my writing. There were several motivating factors:

I realized you should put effort into your passions.

I read Angela Duckworth’s Grit. My main takeaway: Interests are built, not found.

Interests become more fulfilling when you invest in them. This was counterintuitive to me. I always figured hobbies should be casual. I assumed they would stop being fun when they started getting hard. It turns out it’s the opposite. I’ve always said I enjoyed writing but never took it seriously. What happens when I do? Let’s find out!

I decided to write a book

For the better part of a year, I’ve had an idea for a book designed to help engineers increase their product thinking skills so they can deliver better outcomes and have more agency in their work. I hemmed and hawed on whether it was worth it to pursue this idea.

Then, when flying cross country ATL → SFO, I read Write Useful Books. With a new framework in hand, I decided to give it a shot.

Writing a book isn’t just writing a book, it’s writing the promotional copy, the drafts, the pitches, the supplementary marketing material. If I’m going to pull this off, I have to step up my game.

I became an engineering manager

At the start of 2025, I transitioned from software engineer to engineering manager. I view a large part of engineering management as content strategy — Agendas. Specs. Documentation. Slack Messages. Curating information from around the company and distilling it into a digestible format for my team.

Writing is an essential skill for effective management. When you work for a remote company, doubly so.

Examples of generative writing I’ve witnessed

Content blossoming into more content blossoming into even more

I shared a draft table of contents for the book idea with some friends (Great idea from Write Useful Books). I had a tertiary section about “improving your product intuition.” It was the one everyone was most interested in, so it became a sample chapter. Since publishing that, there are a couple of other sections that people found interesting. I can blow those out into entire chapters or larger sections2.

Essay ideas while writing this very piece

Part of my writing practice is posting a short, informal post about whatever is on my mind weekly (At the Edges). When working on this update, I had two other ideas, then realized they probably have enough meat on the bone to be more fully fledged essays.34

Writing creates opportunities

I can’t share too much publicly, but since sharing some sample work from the book, I’ve had some interesting conversations with people I wouldn’t have expected and some exciting opportunities in the “product thinking for engineers” space on the horizon. Those conversations pointed me toward the next part of the book project I should focus on.5

The combination of writing and sharing is powerful. I’m excited to see where this next part of the journey takes me.

I am also terrified. People have bought in and started supporting this “write a book” idea. Now I’m gonna have to sit down and actually finish this damn thing.

Consulting models

While driving to the gym, I was listening to Peter Block's book Flawless Consulting: A Guide to Getting Your Expertise Used. In it, he talks about three models of consulting:

1. Consultant as line worker. They want you to do the work.

2. Consultant as collaborator. They want you to work together on figuring out what to do.

3. Consultant as expert. They want you to tell them what to do.

Consultant failure modes

I realized when an engagement didn't go so well, it was one of two failure modes:

1. I thought I was the expert; they thought I was a line worker. Could you imagine going to Subway, ordering a meatball sub, and the sandwich artist looking at you and saying, "What's the real problem here?" There's a mismatch of expectations, and that conversation isn't going to go well. I realized this is what happened with a lot of my clients. I wanted to be an expert. I wanted to try to solve the big thorny problems, and they wanted a Coder. I ended up writing the code, but we had some awkward meetings, and I wasn't thrilled about it.

2. I was a line worker tasked with an ineffective plan. Who do you think they're going to blame? The internal stakeholder or the outside contractor? Sometimes, it even felt like I was hired strictly as a fall guy to work on a project that they didn't want to do, but deep down, they knew it wouldn’t pan out.

So why now?

So why am I listening to this book now? I haven't done contracting or consulting since 2016, when one of my clients hired me on for a full-time position(I guess that engagement went pretty well.)

The reason is that coaching and consulting aren't titles. They're skills. As I'm getting into engineering management. I see how these types of engagements show up internally, informally, all the time.

When the stakeholders want a report on potential courses of action and the tradeoffs between each, I’m consulting. When I'm helping my reports grow their engineering skills, I'm coaching.

I've realized that management isn't a single skill; It’s a collection of multiple skills. And right now, these are the skills I want to improve.

I think it’s using them to build bespoke utility scripts and tools. LLMs aren’t as effective on large codebases, working with existing code, systems, and conventions they aren’t aware of. AI excels at coding from scratch.

Writing utility scripts is pretty explanatory. With Cursor’s composer, you can provide links to documentation and libraries and knock them out pretty quickly.

Building single function apps with AI

Check out this detailed workflow an engineering manager shared. The tl,dr is:

1. Use ChatGPT1 to help you write a small product spec

2. Use Vite as a framework, put the spec file in a repo, and use Cursor to bootstrap the product

3. Use tinier specs to iterate on smaller facets of the project.

4. Final mile work: manually make minor updates and fixes and add polish.

5. Use the composer to write out GitHub Actions.

6. Ship it!

Using spec files with Cursor is a neat trick. You now have a tool that can read your spec, write your code, and even write a changelog about its process if you ask it to.

It is something I will be trying in the future, no doubt. Not sure what yet, maybe I’ll ask an AI about that to.

Prompts as Scripts

Besides writing scripts, I’ve started thinking of prompts as scripts. Writing a prompt is not a few sentences you throw together once. It can be something you put effort into so that you can use it again and again. I’m moving my prompts into a GitHub repo so I can document them and track changes. I am taking prompt engineering seriously as a discipline2.

A simple prompt framework

• Here’s a rubric I’ve been using when writing prompts. A good prompt includes:
  

• An action to take

• Steps to perform the action

• A persona to emulate

• Examples of inputs and/or outputs

• Context about the action and situation

• Constraints and what not to do

• A template or desired format for the output

It’s not much, but it’s a start. If you have no idea where to begin, I hope that helps.

1. For the past few months, I’ve been working on a book project tentatively titled System is as System Does: The Engineer’s Guide to Product Thinking. Not enough has been said about how engineers have power and responsibility regarding the quality of the software they ship. I want to fix that.

2. I want to share more of the content to get feedback and spark interesting conversations. The first sample chapter is here: How to improve your product intuition.

3. The original plan was to have a separate substack + a separate list for the book content. Turns out this would be more work for me and more confusing for users, so a bad move all around.

Now, I’ve decided to use Substack’s publication sections feature to divide this site into three separate feeds. You can subscribe or unsubscribe to each as you wish1.

1. Root Nodes — The primary feed. Every old article is there. These will be the meat of the sites, articles centering around technology, money, and agency.

2. Derived State — More professional content centered around the topic of the book. If Root Nodes is written by me, then Derived State is written by my worksona: Glenn: Enterprise Edition™️.

3. At The Edges — short, informal weekly dispatches about whatever bullshit I am on at the current moment.

At the Edges ships every(ish) Friday2. The other two will be more sporadic.

The bullshit I’ve been on at the moment is…

Information Architecture

While working on refactoring some internal documentation, I gave How to Make Sense of Any Mess another read. You can read the book online for free at that URL. I wish I could make everyone read this book.

Content isn’t just about the content we create. How we index it, order it, and organize it matters. The organization gives information context and meaning. It tells people what’s important and what isn’t.

Information architecture requires human intent to take content & data and arrange it in a way that makes it understandable. It involves taste and opinion. You cannot offload information architecture to machines.

Search engines treat all content equally. AI is an attempt to outsource understanding, but it's prone to omissions and hallucinations.

So, what am I saying with the new organization of my writing? The purpose of the book project is to encourage developers to take more initiative and ownership when it comes to the work they share with the world.

The theme of this site has trended3 towards encouraging people to take more initiative and ownership in their careers, their finances, and living under a kleptocracy.

My purpose with writing has been to help others overcome resistance. To get out of their own way. To have more agency in doing work that can be proud of in service of getting what they want and helping who they love.

At the end of the day, it’s all one goal; it’s all important to me, so now it’s all one site.