Is Claude Mass Psychosis Affecting Your Home?

Symptoms of Claude Mass Psychosis include:

• The uncontrollable need to tweet hourly about how Claude is “doing 100% of your work” while you sit in a dark room staring at a terminal.

• An unhealthy obsession with Claude Code and the belief that a CLI is your new best friend.

• The persistent, pseudo-philosophical urge to explain to anyone who will listen that Anthropic equity is actually just a call option on all future value within the light cone.

• Running Claude overnight to build complex products that literally no one—not even your mom—is ever going to use.

• Generating massive amounts of slopware that looks like an app but functions like a fever dream.

• Asking Claude to “handle your inbox” and promptly getting phished by an attacker because the model thought a “Limited Time Offer” was a high-priority stakeholder.

• Allowing Claude access to your iMessages, only for it to autonomously text every one of your ex-girlfriends pictures from your 2022 trip to Hawaii with the caption: “Thought you’d like to see these artifacts.”

If you or a loved one is suffering from one or many of these symptoms, you may be entitled to financial or emotional compensation. Contact 1-800-CLAWBAC to submit your claim today. Don’t let the LLM take your dignity along with your API credits.

In All Seriousness

I’ve been seeing a massive surge of Claude-related hype on X (Twitter) over the last few weeks—especially with the rise of CLAWDBOT and Claude Code. Curiosity finally got the better of me. Given the state of the market, I assumed most coding models had hit a plateau—they’re all basically the same, right?

But the discourse made it seem like Claude was different. People weren’t just calling it a helper; they were treating it like a magic tool that could autonomously submit PRs and ship perfect features while you slept.

That was anything but my experience.

Instead of “magic,” I found a deeply frustrating workflow. It was incredibly difficult to see the outputs being produced in real-time, reviewing code changes felt like a chore, and actually building something useful was an uphill battle. For people building real products, I don’t think Claude is that great of a tool. Maybe for someone building a small toy in their garage.

Why I’m Sticking with Cursor

While the “vibe coders” are letting Claude run wild in their terminals, I’ve found myself retreating back to Cursor for almost all of my actual engineering work.

For regular engineering, Cursor just feels better. It doesn’t try to be a “hands-off” agent that deletes your files when you aren’t looking; it stays in the flow of a standard development cycle.

• Real-time Feedback: I can see exactly what is being changed as it happens.

• Granular Review: Reviewing a diff in an IDE is infinitely more productive than squinting at terminal output or waiting for a bot to tell you it’s “finished.”

• Velocity with Control: It makes me significantly faster without stripping away the “engineering” part of the job.

Claude might be the viral star of the moment, but for shipping real code to production, I’ll take the IDE over the psychosis any day.

Finished this book last week and thought others might enjoy it. It seems very self-help-y but I liked the dialog format and also the philosophical underpinning. It’s written as a discussion between a philosopher and student about happiness and social relationships

It was the first time I’ve been exposed to Alfred Adler and his philosophy and framework. That part actually isn’t Japanese at all! Whether you agree or not, a framework that denies trauma, frames all problems as interpersonal problems, and roots happiness in service to community is a fascinating one that aligns with my understanding of the world that I’ve developed in the past 1.5 years.

If you like philosophy, especially in this dialog format, I highly recommend!

I’ve been thinking a lot recently about agentic systems and the shift in how we actually get work done. We’re entering a phase where AI agents aren’t just “chatbots.” They are starting to take over massive chunks of the workflows humans used to own end-to-end.

This shift changes the very foundation of how we build software. For decades, SaaS was built around Systems of Record. But the future? The future belongs to Systems of Action.

The Death of the Discrete Object

When you think about traditional SaaS, it’s centered around discrete objects. Think of a library: you have a book, a genre, and an author. The software is just a way to query those objects. It’s a database with a pretty interface.

There’s a lot of noise right now that Systems of Record are becoming “cheap” to build, and that the cost of this kind of software is going to zero. Maybe it will. But the real reason the old model is fading isn’t just cost; it’s utility. Users don’t just want to store data anymore; they want the data to do something.

The “Domino’s Pizza Tracker” for Work

While a System of Record is centered on objects, a System of Action is centered on timelines.

I like to describe it as the “Domino’s Pizza Tracker” for any unit of work. Whether a human is doing the task or an agent is doing the task, the system tracks the progression of that work in real time.

Behind the scenes, a developer sees this as a graph, specifically a directed semi-cyclic graph. Each node is an action:

• An LLM processing a request.

• A tool call to an external API.

• An agentic loop where the AI self-corrects.

• A “Human-in-the-Loop” moment where a person needs to sign off.

But the end user doesn’t need to see the “spaghetti” of the graph. To them, the “truth” of the system is just a clean, linear timeline of nodes visited. It’s the story of how the job got done.

Why This is the New Moat

This is where it gets fascinating: Systems of Action allow you to build proprietary IP into the execution of the job.

To build a great agentic system, you need incredibly detailed insight into the “Jobs to Be Done.” You are essentially compressing complex human workflows into a digital timeline. By owning the sequence of actions, you are building a deeper level of intelligence into the product than a simple database ever could.

The User Reality

There’s all this noise in the industry right now about “context graphs” and other technical buzzwords. Maybe that matters under the hood, but to the end user, it shouldn’t. The truth is that the system of action should just look like a timeline. It should be a transparent record of discrete events, human-driven or agent-driven, and the ability for the user to see that the whole agentic system is tracking everything.

We’re moving away from software that just sits there and waits for you to type. We’re building software that moves.

I’ve been aware of tools like Wispr Flow for a while (and even looked a bit at building here). This is a hot take, but not everything should be voice. For most short-form, keyboard is easier.

Long form my process has become this:

1. Use apple voice memos to dictate something long (3-10 minutes of speech) into a note

   1. At the beginning of the note, dictate instructions for what the LLM should do with the text (ie write this in my voice, no em-dashes)

2. copy transcript from apple voice memos to my clipboard, paste into my favorite LLM and let it rip

Not everything I’ve written in this time has been this way, but it’s for sure an efficient way to write well-formed posts, todos etc in my voice. The most common use cases I’ve used it for:

• Long form blog posts (2-5 pages)

• Long technical specs for cursor - turning them to .mds or long prompt for the LLM

• Thought out tasks for employees and coworkers

Shorter messages I’m still defaulting to typing - it’s not as much of a time-saver because your keyboard is always right there for up to a few sentences. Voice is a great way though to pound out work while in the car or on the go and have it waiting for you when you’re back at a laptop. Also voice memos are free and already built into your iphone, no expensive subscriptions.

Hope this idea helps someone else!

(ironically this post was not a voice memo)

That CEO had this saying: a 40 watt lightbulb can dimly light a room but a 40 watt laser can burn through steel. The engineer in me questioned the later statement (maybe thick cardboard, but steel is pretty strong), but the principle holds true: focus is what yields results.

The current hype wave around vibecoding reminded me of this idea in the past couple of weeks. Building businesses is actually fairly orthogonal to doing the coding work (which has always been reserved for the nerds), it just happens to be that the coding side of building companies was a hard requirement and used to be a high-bar. Your 10 claude code instances can’t sell the product or talk to customers, for that you need extreme focus, creativity, and resilience.

It’s very clear that there will be more software in the world, and that more broadly agentic harnesses will lead to the commoditization of services as we know them. Building and scaling companies is something very different though that requires creativity, agency, grit, and a lot product insight (and sometimes luck) to make work.

We often look at the invention of the light bulb as a singular moment of technological triumph. But if you really get to the root of it, the story wasn’t just about glass and filaments. It was about the commoditization of light.

Before electrification, light was scarce. It was mostly candlelight or oil lamps. It was dim, expensive, and finite. When the light bulb was invented, people didn’t immediately understand the second and third-order effects. They didn’t foresee neon signs, late-night factory shifts, or the modern internet. But they did know one thing immediately: they no longer needed candles.

Light became so cheap that it ceased to be a luxury and became a commodity. We stopped thinking about the cost of turning on a switch.

We are currently standing at the exact same precipice with AI agents.

The core concept is simple. AI agents are beginning to commoditize what used to be high-margin service industries. We are moving toward a world where services that used to cost hundreds of thousands of dollars will cost tens to hundreds. The end result is inevitable. Services will become ubiquitous and embedded in everything we do, but the margins for the businesses providing them will collapse.

The 10x Deflation

Right now, we are in the boom cycle of the infrastructure build-out. We are obsessing over GPUs and data centers. But as this technology gets deployed, the value of the actual services delivered on top of that infrastructure is going to plummet.

Agents make expensive, human-led knowledge work 10x cheaper and 10x faster.

Consider the current agency model or the vertical SaaS model. These businesses rely on friction. You pay a lawyer a retainer because legal knowledge is gated and time-consuming. You pay a monthly seat fee for software because the utility it provides is scarce.

But what happens when an agent can execute that same workflow for a fraction of a cent in compute costs?

As the technology progresses, we will see a massive compression of value. The service becomes a utility. Over a long enough timeline, industries like software, legal services, and financial planning will stop looking like high-growth tech companies. They will start to look like the business model of a power plant or a water utility.

These will be essential services, required for modern life, but they will not command the premiums they do today.

The Terminal Valuation

This makes the current valuation models for many companies look dangerous. We are used to valuing these businesses based on forward free cash flow, assuming that margins will remain high or expand.

That assumption is wrong.

If the cost of production drops to near zero, the price must eventually follow. The first and second-order effects are easy to see. The current service model is terminal. The current vertical SaaS model is terminal. It will simply be less interesting to build businesses in these spaces because the “moat” of human capital is being filled in by cheap compute.

We are entering an era of service abundance.

The Consumer Surplus

While this is a grim outlook for high-margin service businesses, the end result is actually very good for the end consumer. We are heading toward a golden age of personalized services.

Imagine a world where everything you do online is instantly vetted. You won’t just click “I Agree” on a Terms of Service update. Your personal AI lawyer will read the contract in milliseconds, cross-reference it with your preferences, and ensure your data is safe before you even see the page load.

Your healthcare will move from reactive to proactive. Your personal AI doctor will analyze data from your Whoop band or smart watch in real-time. It will track what you are eating, how you are sleeping, and how you are feeling. It won’t wait for you to schedule an appointment six months out. It will nudge you the moment a metric slips.

This reduces the need for human lawyers and doctors to handle the rote, data-processing layers of their jobs. It pushes those professionals further downstream into more hands-on, high-touch human tasks.

The Rise of the Knowledge Appliance

If raw intelligence is the new electricity, we have to ask who the big winners will be.

When electricity became cheap, the power plants became utilities. The massive value creation didn’t come from selling the electricity itself. It came from the companies that built things you could plug into the wall.

It came from General Electric, Bosch, and Whirlpool.

These companies built the electric motor, the washing machine, and the refrigerator. They took a raw commodity (electricity) and packaged it into a specific, reliable outcome (clean clothes, cold food) that required zero technical skill from the user.

We are about to see the rise of the Full-Stack Agent Company.

These will be the Bosch and GE of the 21st century. They won’t just sell you a “chat” interface or raw access to a model. They will build the Knowledge Appliance.

• The “Legal Appliance”: This won’t be a tool for lawyers. It will be a full-stack agent that handles compliance, taxes, and contracts for small businesses automatically. You plug it into your bank account and email, and it does the work.

• The “Medical Appliance”: A dedicated agentic loop that integrates with your hardware sensors and acts as your primary care physician, handling 99% of diagnostics and triage without a human in the loop.

The winners won’t be the ones selling the power. The winners will be the ones who build the machines that use the power to solve real physical and digital problems.

The New Utility

We used to pay for light by the candle. Now we pay a utility bill for electricity, and we use as much light as we want without thinking.

In the near future, we will look back at the era of billable hours and seat-based pricing the same way we look at whale-oil lamps. Expensive, inefficient, and dimly lit.

The agents are coming. The margins are compressing. And knowledge work is about to become as cheap and accessible as flipping a switch.

One interesting idea with AI is that as software becomes cheaper to create, the limit on how “vertical” you can make this software changes. It trends towards personal software for each professional. We have to recognize that each person has unique needs and each company has unique needs within each role.

I have started to see bits of this along the lines of custom CRMs. One common need for most companies is to have a custom system of record. There are now consultancies that, instead of going to Salesforce, will actually build a custom CRM for your enterprise. What is interesting about this is that it is still at the Enterprise-by-Enterprise level. It is not quite personal yet. However, it is a direct result of the production of software getting cheaper.

Over the past decade and a half, there has been an ecosystem built around vertical software for point categories with the largest companies. They were attacking very large but antiquated markets and building point software for them.

In the future, though, we might see a very different model where agentic software supersedes this. Instead of going to a single vertical platform, the unit economics will make sense for everyone to have a custom solution. Everything from enterprises to small and medium businesses, and even individuals, could choose to have a custom software stack.

There are hints of this in how much traction companies like Lovable and Replit are getting on the custom internal tools front.

I think this is interesting from an entrepreneur’s perspective. Oftentimes it means not competing directly with the large vertical SaaS players. Instead, you go a layer deeper. You get closer to the ICP to truly serve their needs. Agents serve this really well because agents can give a custom experience for each user which is fully unique to them. It fits around their profile of the customer. This creates a better customer experience and also makes your software much stickier.

1. Giving Up (Giving Up)

2. Market Risk (Measure Twice Cut Once: The Idea Maze and Dead Ends)

3. Platform / Partner Risk (Platform / Partner Risk)

4. Cofounder Deadlock (Cofounder Deadlock)

5. Not Shipping (Pivot Hell and Not Shipping)

6. Stagnation (The Stagnation Trap)

7. Raising too much (Raising too Much)

There are more ways, but these are common failures. I might add a bit over time but thought this was a pretty complete list of things-to-avoid. Enjoy!

My birthday comes the day after new years which is just enough time for my friends to get over the hangover. Every year I write down a retro of the last year and goals for the next. Last year I started doing that publicly so this year I thought I’d do that again this year.

Checking in on Last Year’s Goals

Nudge, don’t push

I think I did this ok! Companies are hard but last year I built 4 new products with Excel, got to work with some good friends companies making large impact across the board from high-end institutional crypto to wall street hedge funds to consumer apps, met a bunch of good folks and learned a lot.

Community

I definitely invested in community, both across YC and some old (and new friendships). Grateful for the folks who were in my life in 2025.

Health and Sobriety

Crushed this last year and very grateful to be back in good health. I think I put on something like ~20lbs of muscle while losing 25-30lbs overall which is very positive (old fashion way without GLPs btw just in case you were wondering….). Not drinking alcohol helps a lot plus a bunch of other foundational stuff. Just gonna keep hammering on the consistency next year. I fucked a lot of things up in my life during my mid-twenties not really taking care of myself (no gym, nyc drinking culture) phase but on a good track and excited about it.

v29 Goals

Double Down on Foundation

One of the biggest learnings from last year is how foundation is everything. Without it life sucks, I tend to spend time discontent, unhappy and underperforming my own expectations. Last year I really focused on my foundation and not only was I happier but I also performed much better as a result. This year I’m just going to double down and post more about that process which now includes a diverse set of foundational things like gym, meditation, prayer, and being completely off mind altering substances.

Find time for Mentorship, Teaching, and Reading Every Week

I’ve come to find my experience is quite helpful to younger technologists (or people more generally). I love sharing it by writing and talking to people who want to learn and respond to emails constantly from students, aspiring entrepreneurs, and more to do just that. The secret though is that it isn’t just good for them it’s also good for me because it teaches me too. The reading part is just investing in long-term learning. Hoping to do a lot more of this in my v29.

Bootstrap One Thing to $10k+ MRR

I’ve been hacking constantly on many different ideas working on getting them to stick and this year will be the year I get one to finally get traction (that isn’t a flash in a pan). I’ve been super consistent in my free time on this and not letting off the gas.

Find a Girlfriend

Self explanatory.

Make Friends the Top of My Social Stack Rank

I value time with my good friends so much and hoping that in v29 I can deepen those friendships and double down on some long term friendships that now span decades(!)

One of the strangest things about the startup world is how Venture Capitalists operate. They are constantly chasing what they perceive to be the “outliers”—the massive, once-in-a-decade opportunities.

Oftentimes, founders find themselves in the perfect position to catch this VC lightning in a bottle. Maybe you just graduated from a top-tier accelerator, or you were an early employee at the current “it” company. You have founder-market fit in spades, but you don’t yet have the product-market fit that actually makes a business work.

In hot markets where no one quite knows what the winning formula is, VCs have a mandate to deploy capital. When they see a founder who “looks the part,” they will dump a massive amount of money into the company—often way more than is actually required to build the product.

The Trap: Fundraising Market Fit

When this happens, a founder has two choices. A tiny minority will take the cash, keep the team lean, and quietly hunt for product-market fit while sitting on a war chest.

But the majority of founders fall into a lethal trap. Because they raised a huge sum at a massive valuation, they assume they’ve already won. They think the “market” has spoken. But the market hasn’t spoken—only the VCs have. This is what I call fundraising market fit. The problem with fundraising market fit is simple: VCs are wrong. In fact, they are wrong significantly more often than entrepreneurs are.

The “Saving Face” Spiral

When you let someone else’s intuition—or the hype of a hot market—become the basis for your own confidence, you are almost certain to be wrong about your idea.

Once you’ve taken that much money at a sky-high valuation, you’re locked in. Those VCs now expect you to build a unicorn, and they expect you to keep at it for years. It becomes incredibly difficult to pivot, leave, or find a “successful” exit.

This is a remarkably common way for startups to die. Founders spend years of their lives grinding away on a failing idea just to save face. Instead of looking in the mirror and admitting that the seed round was just a hype wave, they keep the lights on until they burn out.

Raising a ton of money feels like winning, but if you haven’t found the real product-market fit, it’s actually just a great foundation for losing over the long term.

The Two Faces of PMF

When we look at companies that actually find traction, they generally fall into two camps: Strong PMF and Weak PMF.

Strong Product-Market Fit is born from a burning problem. It’s the “hair on fire” scenario. Your product isn’t just a choice; it’s a necessity. The market is practically pulling the product out of your hands. You aren’t “selling” as much as you are trying to keep up with the demand.

Weak Product-Market Fit, on the other hand, is built on “nice-to-haves.” It’s what happens when you have a world-class, resilient, and incredibly talented team. They’re smart enough to build a great tool and gritty enough to force it into the hands of customers. They get traction through sheer force of will.

Here’s the reality: the vast majority of products on the market today have either no PMF or Weak PMF. And let’s be clear—Weak PMF is perfectly fine if you’re running a bootstrapped business, a small agency, or a steady lifestyle company. But if you’ve signed up for the venture-backed path? Weak PMF is a death sentence.

The Stagnation Cycle

One of the most common failure modes I see among seed-stage founders is what I call Stagnation. It starts with a “good” idea. The product works. The customers like it. The company makes money and might even be profitable. But—and this is the crucial part—it doesn’t grow at the rate necessary to sustain a venture-backed trajectory.

These founders get trapped. Because the business isn’t “failing” in the traditional sense, they spend decades working on it. They keep the team lean. They stay small. They grind for ten, fifteen years only to eventually exit for a modest amount that barely clears the liquidation preference.

They never get that runaway success they set out for. They spent their most productive years tending a garden when they meant to build an empire.

The Misalignment Nightmare

The real tragedy of stagnation isn’t just the slow growth—it’s the misalignment.

When you raise venture capital, you are making a pact. You are promising a specific type of return in exchange for fuel. When you stagnate, you end up with a cap table full of investors, board members, and stakeholders who are unhappy with the outcome.

You’re profitable enough to stay alive, but too slow to ever provide a venture-scale exit. You’re stuck in no-man’s-land. The founders end up building something that no one is truly happy with—not the investors, and eventually, not even the founders themselves. It usually ends one of two ways: the company is eventually shut down out of exhaustion, or it’s sold for scraps just to get it off the books.

Why It’s Better to Fail Fast

This might sound counterintuitive, but in the world of startups, it is actually better to fail fast than to stagnate slowly.

When you fail fast, the lesson is sharp and immediate. You realize the market isn’t there, or the problem isn’t burning enough, and you move on. You preserve your most valuable asset: your time. Failing fast gives you the opportunity to take more “shots on goal.”

Stagnation, however, steals your time. It’s a “slow-motion” failure that tricks you into thinking you’re winning because the lights are still on. It leads people to spend years of their lives building something they don’t quite believe in, or something they know, deep down, won’t ever truly “work” at scale.

The Bottom Line

If you’re raising venture capital, you have to be brutally honest with yourself about the type of PMF you’re chasing.

Don’t get caught in the trap of “good enough.” Don’t let a talented team mask a weak market. If you find yourself stagnating, have the courage to pivot or walk away. Because the only thing worse than a startup that fails in a year is a startup that takes ten years to tell you it was never going to make it.

Pivot Hell is that Limbo where a team hasn’t found an idea they actually have conviction in. They are moving, but they aren’t going anywhere. The primary reason they stay stuck? They stop shipping.

If you’re currently spinning your wheels, here is how you find your way out.

The Brutal Math of Ideas

Let’s be real: 99% of ideas are bad. Most things simply will not work. Only about 0.01% of ideas have the potential to become massive, era-defining companies. To hit that scale, you need the trifecta:

1. A Good Idea

2. A Good Market

3. A Good Team

If you have the team, the question becomes: how do you actually land on the idea? Standard startup advice tells you to pick something you’re interested in—Founder-Market Fit. That’s fine for identifying useful problems, but it carries a massive risk: Founders fall in love with their first idea and double down on a loser.

Why Pivot Hell is (Actually) a Gift

Believe it or not, Pivot Hell is a necessary phase for most startups. It’s where your team forms its intuition.

It’s the period where you realize what doesn’t work, which is just as important as finding what does. But there’s a trick to getting out of it. To exit Pivot Hell, you need conviction—the kind of conviction that makes you comfortable spending the next ten years of your life on a single problem.

That is a incredibly high bar. You can’t reach it by thinking; you can only reach it by doing.

The Only Way Out is Shipping

The most common mistake I see teams make in Pivot Hell is that they stop shipping. They spend quarters—sometimes years—theorizing, researching, and “pivoting” on paper.

If you want conviction, you need a product live. Period.

You need to see customers use it. You need to feel their response.

• If it’s “kind of nice”: That’s not the pain you want to solve.

• If it’s a “burning desire”: They will slap together the most half-assed version of your product just to make it work for them.

That “slap-test” is your signal. If a customer is willing to use a broken, ugly, incomplete tool to solve their problem, you’ve found it. That’s where conviction is born.

Set Your Criteria

Don’t let Pivot Hell become your permanent residence. Founders give up here because they don’t have a “success” metric.

1. Define “Working”: Set a hard criteria for what traction looks like.

2. Get Uncomfortable: Get out of your comfort zone and hand customers those “half-assed” products.

3. Listen to the Friction: If they aren’t fighting to use it, move on to the next experiment.

Stop trying to find the perfect idea in your head. Build something, ship it, and let the market tell you if you’re right.

Inherently building new things requires undue amounts of sacrifice of something bigger, whether it’s a company, project, or idea.

At the beginning of Covid I read one of my favorite books by Viktor Frankel called Mans Search For Meaning. It’s about survival during the holocaust from the perspective of a psychologist. The common thread that made people more resilient is that survival was at the service of something bigger themselves, whether it was their family or some broader purpose.

I think the same goes for many normal situations in life, and those of us who can recenter on that purpose are the ones who have the ability to create the things that power us into the future.

It’s different for everyone, but within this idea are forms of spiritual growth that are key to the insights that will help us create the future for our children.

Here’s a link to that book for anyone interested in reading. It’s one of my all-time favorites

The playbook is simple: A platform—whether it’s the App Store, a browser, an OS, or a blockchain—needs to grow. But the owners of these platforms don’t actually know what the “winning” applications will be. They have the distribution, but they don’t have the crystal ball.

So, they open the gates.

The Flywheel Trap

They make the platform “developer-friendly.” They offer APIs, documentation, and a seat at the table. They want you to build there because every successful app you build adds value to their ecosystem. You are the R&D department they don’t have to pay for.

Think about the early iPhone. It had flashlight apps and quirky games. No one at Apple fully realized that music streaming or ride-sharing would eventually become the core utility of the device. They let the market figure it out.

But here’s the reality: Every platform has an owner. And every owner is a gatekeeper.

From Partner to Predator

The incentive structure changes the moment you find product-market fit. Once an application like Spotify, Pandora, or YouTube Music proves that streaming is a massive, high-margin business, the platform owner (Apple) has a distinct incentive to take that market back.

They don’t just watch you succeed; they move to compete. And when they do, you aren’t playing on a level playing field.

• They control the economics: They take a 30% cut of your revenue while their own version (Apple Music) keeps 100%.

• They control the real estate: They can pre-install their app or bury yours in search results.

• They own the rails: They can change the rules of the game at any time.

The Cost of Being the Lab Rat

I’ve seen this happen over and over. I’ve had my own companies killed this way. On the outside, the platform owner looks like your biggest cheerleader. They want your traction to build their flywheel. But as soon as you get successful enough to matter, they will move to clone your product or build a “better” integrated version to reclaim the market.

The Lesson: When you’re thinking about your strategy, you have to look past the initial “friendliness” of a platform. You need a deep insight into the platform’s own incentives.

If your entire business model is a feature that the platform owner can eventually integrate into the OS, you aren’t building a company—you’re just doing free market research for a giant.

Foundation is everything

Your personal foundation is the foundation you build the company on. If you fail (in life) the company you’re working on will surely fail. Therefore it is important to put your own personal foundation before anything to do with your company if you want to prevent your company from failing.

Everything is a power law

The engineers you hire, customer contract sizes you close, days of crazy high or depressingly low growth, the people you date, the insights you get from the books you read, the level of fitness you have, the food you eat, company outcomes, how hard you and your team work, the personality and talent of the sales people you hire, the attention each launch gets and more. Everything is a power law. Your job is to find the outliers.

Winning

Winning is what makes people happy

People are happy when they’re winning and sad / frustrated when they lose. Startups have a lot of losing so it’s important to focus on winning.

When making deals make sure the people you want around win

The people who win with you are the people who will stick around. Make sure that the people you value win when you win and they’ll keep playing the game with you.

Resilience

Faith

You can’t have resilience without faith, anything else is based on ego which is bound to fail.

Pain Tolerance

The easiest way to find outliers is to expose yourself to a lot of not-outliers so that you know one when you find one. This will mean losing a lot in order to learn how to win.

Skin in the game is the foundation of growth

You need to take risks in order to learn the lessons that will be the foundation for your growth. You can read, listen and learn all you want from others but without skin in the game you’ll never make progress.

What matters for building a business (in order): team, market, idea

The team building the business are the ones who decide everything else so they matter more than everything. Second to that is choosing a market which will determine how fast the business can grow. Third is the idea within that market and if the customers truly want it.

Trust your intuition

You as a leader have the most information on any given situation to make the high-level decisions the business needs. Anyone else presenting with strong opinions has bias and blind spots that almost always make those decisions the wrong ones.

The best products are more like paintings than they are tools

The best product builders are closer to artists than they are engineers, so if you want to create great products focus on learning how to paint (build) and then making great art (creating products).

Do what you love

If you don’t love it there is no shot that you’re going to build something that is an outlier because it’ll be too hard and you’ll give up. Focus on what you love first.

The future hinges on doing the next right thing

In every moment each one of us has the choice to help determine the future by doing the next right thing. Sometimes those things are non obvious, sometimes they are obvious but if you can manage to keep doing the next right thing over and over again that is one of the main keys to doing something great.

Then the question comes up, what makes founders give up?

Most startups are long games where the core of the problem is just staying in the game long enough to see yourself succeed. At a certain point almost every one becomes irrational to keep going under certain circumstances unless you love the work, customers, and idea. Founders run out of energy far more often than they run out of resources.

The core to this answer is running out of emotional energy, bandwidth and resources to keep working on an idea (Note: this does not mean running out of money_ I’ve been around hundreds of companies and I’ve seen less than a handful die because they ran out of money.

So, how do you not lose your energy?

There are a bunch of answers to this question, but the most obvious is have a resilient mindset to things not working for much longer than you’d expect. It’s very uncomfortable for people who spend their careers working on things already established to admit that they don’t know if the thing they’re building today has legs.

The most important founder trait is thriving in that chaos, and allowing yourself the time and space to make good decisions that will help you get out of it. It’s like jumping into the deep end of a swimming pool. Better learn to tread water quick!

Aren’t there valid reasons for giving up?

Absolutely yes, there are many. Sometimes founders can’t keep going for personal reasons, markets don’t pan out, teams split, or major incidents threaten businesses. These happen all of the time, but your mindset in those moments will determine if you can keep going or not. Sometimes it’s reasonable to shut down, I’ve seen founders threatened, family get sick, or worse. Valid. But in those moments of crisis your mindset, energy and willpower will determine the future of the thing you’re building.

Choosing a cofounder is one of the most important steps in starting a company (and yes you can do it alone, which is hard in other ways), and most people do it haphazardly. They say “I liked working with this person at xyz company“ or “this person is my best friend of course we should do this together“ and shoot from the hip. The problem with choosing cofounders this way is that you never really get to see them in those stressful, make-or-break moments prior to being in the first one together and oftentimes that first moment is one of seemingly life or death.

There are two core questions when working with a cofounder: do we both understand the risks (ie is doing a startup good for our lives right now) and how do we tie break decisions?

The risk question is really more one of “do you really want to do this?“ which is surprisingly tricky. Different people have different motivations for starting companies. Most often though folks won’t be honest with themselves that it’s ego driven, not driven from a place for passion for a problem or market. I’ve written about finding cofounders, but one of the easiest ways to find a good cofounder is to just go build something and see who tags along and is useful. You can also recruit and trial (before incorporating) but that is a longer process that requires more resources.

The second question is: how do we tie break decisions? I see sooooo many companies die in this mode where cofounders can agree on something that sticks. You need to have a tie break. Many VCs will advocate for 50/50 splits between cofounders, and while this can be a good thing for VCs, it’s often not the best for the founders themselves because when the going gets tough and we need to decide some major choice where there’s deadlock how do we decide?

A pretty common failure mode is choosing to pivot or double down. This is always a stressful time because both cofounders know what we have isn’t working but they don’t exactly know how so there’s a game of convincing the other to take another path (which also might not work). In these moments you need a framework for who breaks the tie.

Good read below on cofounder roommate agreements below that can help give you a framework for decision making:

The best and ultimate way though is to either have an odd number of cofounders (natural tie break) or a single voting share that tie breaks at the corporate level. This way it’s clear who gets the final call. Be aware, in the case of social breakup between cofounders (this happens) then the person with the voting share is the one who stays and the person without it goes (just math). What I’ve found is that in those situations 50/50 can sometimes be better because if the person with the voting share doesn’t have the skills / network / talent to take the company forward the company just dies by default (and this is why VCs like the 50/50 split because it often means the “key man (person)“ stays.

Hopefully this is insightful, and helps someone set up a cofounder relationship right. The main insight I have is that you shouldn’t rush into cofounding. Date a little first before getting married so that you’re sure you’re getting married to the right person.

1. Lose close maternal grandparent in August, speak at funeral

2. Destabilizing political violence and regular violence (national and local)

3. Travel - get Covid

4. Start new job (exciting)

5. Travel - get cold

6. New job ramp up (which is admittedly fun)

7. Broad FUD around world economy reordering based on instability the technology sector (mine)

8. Travel - get flu

9. <now we’re here>

I’m gassed. I’m pretty sure I’ve had every illness that I could get at this point, so at least we should be set on that front. It was already pretty miserable as of October. I guess the other stuff could get worse, but hopefully won’t and at least I can be grateful for the work I’m doing, my dog, family and friends. Reorienting on that helps.

Sometimes, if you’re ambitious like I am, it can be easy to stay focused on the end goal. One thing that can be hard to see in the moment is that the thing that helps you get there the best is actually taking a step back and reorienting and getting back to the foundation. I wrote this post over the summer about how breaks are for losers and one of the core parts of that is working as hard on yourself as you do on other things. Couldn’t be more true now.

Funny thing about these slides being ~10 years out of date is you can see the treasure shifts - iTunes, maybe not today but Spotify and Apple Music are big businesses. Some of these even I have not heard of (like KaZoA wtf is that lol).

Balaji paired this concept with a concept of “verbs“ - things that you can do with money (buy/sell, stake, purchase, send, earn). Each verb had a specific product vertical that we were attacking at Coinbase. Now 7 years after that talk, I’m starting to wonder where the treasure exists.

One core question I’ve asked myself over the last few years is this: is the decentralized money on the internet maze just filled with dead ends? I’ve spent (too much) time here and can’t for the life of me think of a winning strategy long-term besides Bitcoin as digital gold. What if instead of a maze filled with centralized custody / decentralized custody, lending, exchange, transaction (b2b vs b2c) over a long enough time horizon it actually looks like this with only two options:

I’ve also been thinking about this with AI, and with a slight amount of future-sight think the beginning of the maze might look something like the following:

This still has to play out, but what I’m realizing is that as your ideating having this maze mapped out (literally) is quite important. It’s important on a per-builder basis and on a team basis to really try to find root truth to why something will work or not and is also where second mover advantage with some lessons from the first might pay back handsomely.

As you zoom out even the future of software companies might look something like:

What you might realize with these slides is that you also need to think through first what your foundational decisions are that are make or break and that can’t change - make those product + strategy decisions first, then execute.

The idea maze is totally yours to own, but choose wisely because many things die two years before shutdown because the wrong decisions were made at the very beginning.

My morning routine has become very simple over the past 6 months (no ice baths). It’s this: be grateful.

Today I’m grateful for the journey. It’s an incredible experience to get to do what I do every day and even in the middle of the ups and downs that gratitude is a constant. Keeps me grounded into whatever challenges we face.