Two attacks in 48 hours on the most visible person in AI. Someone who has undeniably changed the world.

I keep walking around San Francisco and the energy is different in just two months. People lower their voices at dinner when the topic turns to work. They hedge when they talk about their companies. Everyone has some private number in their head about what they are worth if everything changes in the next eighteen months.

A founder I angel invested in has not updated his board in a few months. His product works, revenue is growing, but a new Claude feature does 80% of what he spent two years building. He does not know what to tell them. So, he has resorted to saying nothing.

YC founders refresh Anthropic’s changelog the way traders watch earnings calls. Every release is a potential kill shot. Anthropic added $11B in annualized revenue in the last month and OpenAI just closed $122B at an $852B valuation. These companies ship features monthly that erase startups overnight.

School admissions decisions landed last week. Parents I know who got lucky are already past the celebration and onto the next panic: can we still afford to live here in three years? Anthropic, OpenAI, maybe others will IPO soon. That money floods a housing market that is already broken. You are either on the rocket or watching it from the sidewalk.

Layoffs arrive in quiet monthly waves now. No headlines, just Slack messages. “Restructuring.” Everyone knows what it means. The agents are now doing what teams of people used to.

A recent NBC poll found Americans view AI less favorably than ICE. The guy who threw that Molotov cocktail was 20 years old. He was not angry about economics. He was simply terrified of extinction.

And yet the laptops in the Mission are still open on any given Tuesday. Agents are running overnight but the people who launched them are already onto the next problem by morning. The founder with the upside down cap table called me yesterday. He wanted to pitch his next idea.

The fear is the same everywhere. The 20-year-old with the Molotov cocktail and the founder ripping apart his roadmap at midnight are living through the same moment. One of them reached for a bomb. The other is optimistically thinking about what he wants to build next.

Build, don’t burn. That is the only way through this.

Comparison is how ambitious people procrastinate. It feels productive because it involves the same subject matter, the same industry, the same people. You think you are studying the game. “Benchmarking”. Meanwhile your actual work stays untouched.

Tracking competitors used to make sense. Who shipped the feature first. Who priced lower. Who hired the sales rep you wanted. Execution was hard, the race was about execution, and watching was a legitimate part of the playbook.

AI killed that logic. Forty minutes vs. six weeks. Anyone can build the obvious thing now, the market map, the landing page, the MVP that does what the customer said they wanted. When execution is free, there’s nothing left worth copying. The surface is drowning in adequate, and the playbook everyone’s studying is already a commodity.

What wins in the near term are secrets. Things you can only learn by going somewhere nobody else thought to look, a distribution channel that only works because you understand a specific buyer’s workflow, a technical decision that seems wrong until you’ve spent months inside the problem.

I see this split in pitches every week. One founder has the competitor slide memorized. Knows the Gartner quadrant. Can recite every feature comparison. I ask why Customer #3 almost churned and they blink. No depth, just the standard “churn’s within benchmark ranges.” They know the map but they’ve never been to the territory.

Then there’s the founder who interrupts my question because she’s been thinking about it for 400 hours and already knows where I’m going. Has a take on some pricing quirk in construction bidding that sounds obscure, shouldn’t matter, and turns out to be the main insight. She found a crevice nobody else thought to enter and just kept going.

I ask about competitors and she stares like I asked about the weather. She’s somewhere they can’t see, and she got there by chasing questions everyone else dismissed. Don’t get me wrong. She’s competition-aware, but she doesn’t consider them in the same game.

The founders still tracking competitors are playing a finished game. The best companies are icebergs, and you can’t build one by watching what others ship.

Even investors should not be immune. I stopped reading the same ten newsletters everyone else reads. Started cloning repos instead, deconstructing and reconstructing frameworks I didn’t know because I wanted to understand how they actually worked before the consensus formed. Learned more about where AI tooling is heading in three hours - much more fun than hearing the same three tropes from peers.

There will always be a superficial leaderboard. Funding rounds, feature launches, press coverage. Don’t let it define you. Play your own game.

“You’ve already made your decision. You just want me to validate it, right?”

Silence. Then a laugh. He’d been working on this thing nights and weekends for over a year. He had a family, good job, kept telling himself it was a hobby. But he just couldn’t stop. Every morning he woke up thinking about it. Every meeting at work felt like time stolen from the real thing.

That’s the pull.

You can’t justify it on a spreadsheet. You can’t even explain it to yourself half the time. But it won’t go away. Especially when you try to make it.

Lenny Rachitsky has 1.2 million newsletter subscribers now. Before any of that, he was on a trip to Joshua Tree, on psychedelics, sitting on a rock for three hours watching some kind of Buddha visualization with one phrase stuck on repeat: I have wisdom to share. Three hours on one rock with one phrase.

Nine months into writing weekly, he told a friend he didn’t understand why he kept doing it. That he should be focusing on his startup. His friend asked him something he couldn’t answer: how often do you enjoy something and people actually value it? Maybe follow that thread.

He almost didn’t. Most of us almost never do.

Everyone’s felt some version of this. A problem that nags at you. A market that looks broken in a way only you seem to notice. A skill you keep building even though nobody asked. If you can’t articulate the thing, how can it be real?

So, you wait. The timing isn’t right. You need more experience. After the next promotion. After the next vest. After the kids are older. Your family, your friends, they reinforce it because they love you and stability feels like the loving answer.

But, that thing keeping you up at night? The problem you think is so glaring someone smarter must already be on it? There’s a very good chance nobody is. The gap between what you see and what everyone else sees is not a reason to wait. Act on it. Most of my greatest mistakes in my career have occurred because of this gap.

The cost of ignoring the pull is time. Scrolling past someone else’s launch in three years thinking, I saw that. I had that idea. Nobody is ever haunted by the things they tried.

The pull doesn’t show up often. FOMO shows up daily. Impulse shows up daily. The pull lingers for months, years sometimes, and you can’t argue your way out of it no matter how hard you try.

If you’re feeling it right now, you already know what it’s about. Share it openly, invest in it, build something based on it. Today.

In one, founders are swapping tools by the week. They max out token limits before lunch and feel anxious about it. A solo engineer is now shipping what took a team of five last year. People refactor workflows every quarter and still feel behind. Everyone around me has welcomed AI into every part of their workflow (thinking AND shipping).

In the other, I sit across from Fortune 500 execs who still think of LLMs as a toy. VPs of Engineering who haven’t opened Cursor. And, these aren’t just random people on the street. These are people with decades of experience who have refused to update their mental models. Their priors froze when LLMs couldn’t count the number of Rs in strawberry. The models have improved every week since. The harnesses around them have gotten dramatically better. They wouldn’t know because they refuse to get their hands dirty.

It’d be one thing if these realities existed in different parts of the world. But, I’m seeing them in one place. At the same dinner.

Block cut 4,000 people last month. 40% of the company. Jack was blunt: smaller teams with AI tools do more and do it better. He predicted most companies will follow within a year. The stock jumped 24%.

Working at startups, the “10x engineer” was always a mythical being. That one person on every team who shipped while everyone else was still debating the right abstraction layer. We told stories about them like Usain Bolt. Aspirational. One per generation.

I think the world has not priced in the fact that AI has raised the floor. A human paired with AI can already perform at the 10x level. The mythical 10x didn’t get better. The floor just rose to meet them.

The talent market used to follow a bell curve. Most people clustered around average. Companies built compensation and leveling around that distribution. Show up, do solid work, get promoted on schedule. That curve is splitting in half. Every week the tools get more capable, and more people will be expected to perform at the new level.

The people who have truly embraced AI have figured out their job is to be a conductor. Hold all the context, the feeling, the texture in their head while using a fleet of agents to handle the execution. It’s like playing chess while making sure a dozen trains don’t run into each other, even though each train has its own conductor. You just need to hear when something’s off.

The other side still meets their OKRs. I was at a dinner recently and sat next to a VP of Product. 16 years at big tech companies. Runs a tight ship. When I asked what he’d personally built with AI, he listed things his team had done. His hands hadn’t been dirty in years. Built a career on reliable competence. Shows up, hits deadlines, keeps the trains running.

That profile used to be the most valued in corporate America for half a century. $300K and a comfortable trajectory. The tools just caught up to it. Heck, let’s be honest, they are better than them.

AI should be renamed amplification intelligence. It simply showed who already had agency and curiosity, and amplified them. The variance used to always exist. Now, it’s clear as night and day.

This is coming for all of us faster than any of us like to admit. The two realities are colliding fast. And when they do, only one version of “good enough” survives.

Software companies borrowed the whole model. PM writes the spec, designer mocks it up, engineer builds it, QA tests it, marketing launches it. Stations on a line. The org chart grew around those skill gaps like a city grows around a river. Agile tried to fix it. We moved from waterfall to sprints but kept all the handoffs. Faster cycles, but the same stations.

AI collapsed those skill gaps in months. The org chart stayed exactly where it was.

I’ve been a generalist my entire career. Never picked a lane. At every company I just found the most important problem and went after it, regardless of whose job it was supposed to be. For most of my career that felt like swimming against the current.

Not anymore. No fast-growing startup I know hires just a PM or just a marketer anymore. They want tinkerers. That marketer? She’s in Claude figuring out what the last deploy changed, building the landing page herself, writing ad copy, and launching the campaign before she goes home. No tickets. No handoffs. The engineer ships the feature, creates the launch video, spins up agents to review the PR, and owns what happens after it goes live. Nobody told them to work this way. There was just a problem and they went after it.

Early startups always operated like this. No boundaries, just outcomes. The difference now is that AI lets you keep that scrappiness at 50, 100, 200 people. The gaps that forced specialization at scale are gone. These teams pay more in tokens than they’d spend on new hires.

I was at the office with a founder in our portfolio yesterday. He walked me through a feature. Customer research, design, code, launch. All him. I asked who else was involved and he kind of laughed. “Why would I hand it off? I had all the context.”

“Pick a lane” needs to be demolished. What to build is way more important than how to build now. Generalist vs. specialist was always the wrong debate. Ownership versus dependency. That’s the only question that matters.

Most people don’t cross though. I get it. Your title starts to feel like who you are. Touching someone else’s territory feels wrong. And yeah, the first few times you’ll be bad at it. But a rough version you ship with full context in your head beats whatever gets delivered six weeks later. Three handoffs strip out everything that matters. Nobody gives you permission to cross. Performance reviews actively punish it. We need to stop obeying a system that stopped making sense years ago.

Companies that don’t fix this will bleed their best people to ones that already have. Give them space, ambition, hard targets and watch what happens.

Ford’s assembly line was genius for 1913. The resistance between stations was real.

That resistance is gone. And every day you stay at yours, someone without your title, your permission, your experience is shipping the thing you’ve been waiting to hand off.

Economists call this revealed preferences. The gap between what you say and what you do. Every platform before TikTok asked users to build their own feed. Follow accounts, like posts, tell us what you like. TikTok skipped the asking. Dwell time over likes. Subtle, but important difference.

I’ve been thinking about how modern LLMs learn. There’s this technique called RLHF (reinforcement learning from human feedback). Human raters look at two outputs and pick which one’s better. Thumbs up, thumbs down, repeat a few million times. That’s how ChatGPT, Claude, Gemini all got less weird.

The problem is RLHF (typically) trains one model for everyone. Millions of preferences blended into one reward function. Your thumbs up and mine averaged together. The model converges toward what most people like most of the time.

What passes for personalization today doesn’t change this. Memory stores what you told the system. Your name, your job, your tone. Stated preference bolted onto the same model everyone else uses.

Tab, tab, tab. Every software tool throws off micro-decisions. Accept, reject, edit, regenerate, abandon. Code completions you take versus skip. Email drafts you send versus rewrite. Each one is a vote.

Unconscious, mostly. After a while you stop noticing. 2 AM, accepting code suggestions because you have token anxiety. Third rewrite on that email because your boss might read it wrong. Training data is the last thing on your mind.

Feeds wish they had data this clean. Scrolling is semi-conscious. You know you’re being fed content. But autocomplete? You don’t perform for autocomplete.

Right now, this data improves models for everyone. Cursor uses your accepts and rejects to make Cursor better for all users. Convergence.

It doesn’t have to work that way.

Imagine a version where the data makes your model diverge from mine. A model that becomes irreversibly yours. Your “AI” and my “AI” turn into different products over time because we used them differently, and the system noticed.

Last year I wrote about forking your tools. The power to consciously remix an interface. You decide to fork. Here the product forks itself. You didn’t configure anything. You worked, and your patterns became the product’s patterns.

The moat is the divergence, not the model.

Switching costs transform. Files travel fine between tools. Export, import, done. But the understanding doesn’t transfer. Thousands of micro-preferences the system learned that you never articulated. Switching means teaching a new tool from scratch.

I probably wouldn’t bother.

The infrastructure for per-user divergence isn’t here yet. But it’s coming. Capture the behavioral data now. Log the accepts, rejects, edits, regenerations per user. The companies sitting on that history will have a head start that takes years to replicate.

Users will feel it the moment they try to switch.

On the flip side, I talked to an exec who was proud of his AI rollout. Every team had copilots, dashboards tracked adoption, and he walked me through efficiency gains for twenty minutes. I asked how his hiring plan had changed. It hadn’t. He couldn’t name a single metric that moved because of AI, and headcount was still going up. He’d given every person in every role an AI assistant and left the structure exactly where it was.

Same dependencies between product and design. Same review cycles between design and eng. Same meetings with eight people so everyone “has context.” Product still says “prioritize.” Eng still says “capacity.” The board wants Q2 projections and something about “product velocity.” Everyone’s busy. They look around the room and it feels like progress.

These roles used to make sense. The PM had the idea, the engineer built it. That separation existed because you could not do the next step yourself. It was a real skill constraint, not some process thing. AI blew that up and nobody seems to have noticed, or at least nobody redrew the org chart.

At an event for non-technical folks using Claude Code, I watched an investment banker build a shopping app and push it live for customers that same afternoon. The distance between idea and done is basically zero now, for everyone, and most companies are still organized around a distance that no longer exists.

No fast-growing startup I know is hiring just a PM or just a marketer anymore. They want tinkerers. The person who talks to the customer in the morning, builds the fix by noon, and ships it before dinner. One brain holding the whole problem, start to finish. Not because they can’t afford to specialize. Because every time work changes hands, context leaks. And they learned to just not have that problem.

The best AI-native startups I see are spending thousands a month on token costs instead of adding headcount. Their biggest line item isn’t salaries. It’s compute. I think that says more about where companies are headed than any AI “strategy” deck.

Your company still has a role for every function and a gap between each one. Those gaps add up to weeks, and in those weeks your customers are already talking to someone who doesn’t work that way.

Every Fortune 500 is hiring a Chief AI Officer right now. Same instinct as “VP of Innovation” a decade ago. You created a role so the rest of the org could stay exactly as it is. Tobi Lutke doesn’t have a CAIO. He is the CAIO. His conviction comes from building, not briefings. Most CEOs would rather hire someone to have that conviction on their behalf, and it shows.

Startups have always been faster, that’s not news. What changed is that AI lets every person on a team take something from idea to shipped product without needing anyone else to touch it. Per-seat incumbents have the same tools on every laptop and a structure that forces the separation anyway.

The pot keeps boiling.

Nobody questions it anymore. Half the room is thinking the same thing. The other half are probably checking the progress of their agents. At a party.

All the parties are sober now. Young people don’t drink because they’re going back to work after. Not inspired by Bryan Johnson, although that’s probably a factor. The buzz they want now runs on tokens per day.

I keep noticing it on walks through the Mission. Laptops glowing everywhere. Cafes, sidewalks, heck even park benches. People walking with screens open like a flashlight guiding them somewhere. Less drunk laughter on the streets these days. More keystrokes.

Dinner conversations used to start with “what are you building?” That’s over. Now it’s “how many agents do you have running?” People drop the number the way they used to drop their follower count. Quietly competitive. The flex isn’t what you’ve accomplished anymore. It’s what’s working while you’re sitting here not working.

The vocabulary is what really gets me though. People describe models the way sommeliers describe wine. This one has better taste. That one hallucinates with more confidence. Opus is bold, Codex is smooth. They talk about harnesses and reins like they’re controlling horses. Invisible whips directing invisible labor. Someone at a dinner said they keep “Claude on a tight leash for code review but give it more slack for creative work.” We’ve started borrowing the language of how we treat animals for something none of us actually understand yet.

Waking up and checking what your agents produced overnight is the first thing now. Before coffee. Before texts. You open your laptop and grade homework you assigned in your sleep. Some of it is good. Most needs rework. But you start shipping a plan before you sleep just so you can wake up to more code written overnight. Saturdays became uninterrupted build windows. No meetings, no Slack, twelve hours of you and your agents. Sunday morning X is all terminal screenshots and shipping receipts. “What’d you ship this weekend?” replaced “what’d you do this weekend?”

The anxiety is rational, which is why it sticks. Every week some new benchmark drops that makes last month’s workflow feel prehistoric. Codex ships overnight processing. Opus gets faster. Context windows double. None of it reduces the pressure. It multiplies it. You can do more now. And someone already is. The window to be first at anything feels like it’s shrinking by the day. Literally, by the day.

I replaced Netflix with Claude Code. I lie in bed thinking about what I can spin up before I fall asleep, what can run while I’m unconscious. Reading a novel feels indulgent now. Watching a movie without a laptop open feels wasteful. This voice in my head that says “something could be running right now” just doesn’t shut off. I’m not even building a company. I’m just addicted to building my random ideas.

Everyone here knows they should step away more. That’s not the problem. The problem is what your brain does when you try. I still take aimless walks. The agents come with me now.

Last week, a Slack popped up. “Exciting news.” Your CEO is joining Microsoft. Most of the team is going with him. You are not on the list. You get paid, but the mission leaders have abandoned you.

In 2022, Adobe tried to buy Figma for $20 billion. Regulators decided to block it so Big Tech had to find a side door.

Microsoft paid Inflection $650 million in “licensing fees” and hired the CEO and most of the 70-person team. Amazon did the same with Adept, taking 80% of the technical staff. Google hired back the Character.AI founders for $2.7 billion, then repeated it with Windsurf and Hume. Same structure every time: license the technology, hire who you want, leave the rest.

Not an acquisition. No regulatory review. Everyone who built the thing doesn’t come along.

Now, founders are leaving their own companies. Last month, two co-founders of a $12 billion startup went back to the previous company. A year ago, a CEO left his $32 billion company for a competitor. When the people who started it don’t stay, why would anyone else?

In 2012, Instagram sold for $1 billion. All 13 employees joined Facebook. Founders stayed for six years. Everyone who took the risk shared the outcome.

There was an oath. Never written down. No founder signed it. No VC swore it in front of witnesses. But everyone understood: you do right by the people who bet on you. Employees who took pay cuts. Investors who wrote checks when nothing was proven. Partners who turned down safer options.

That oath is dead.

The week after your CEO announces, everything looks normal. Same standups. Same Slack channels. The people who got picked are negotiating their equity packages. Everyone else is quietly updating LinkedIn.

Why take the risk when the best case is watching your CEO leave without you?

VCs who funded your startup learned the same lesson. Five years ago, backing a competitor to a portfolio company was taboo. One fund forfeited a $21 million stake rather than create a conflict. Now mega funds back three, four, five companies in the same category. Fund sizes doubled. Companies stay private for 15 years. They can’t afford to miss a large category. So they bet on everyone and let the market decide. Watch them squeamishly talk about how “each” one is different.

I work at a boutique fund. We can’t hedge like that. We have to pick. I’m biased but that’s not the point.

The point is what we’re teaching. Every employee who got that Slack learns that “mission” is a recruiting pitch. Every founder who watches their VC fund a competitor learns they were a bet, not a partner.

I still back founders. I still believe there’s never been a better time to build. But I try to find people who still believe the oath matters. The ones who would rather go slower than abandon the people who believed in them. Even when faster often means richer.

So pick carefully. Ask your VC who else they have backed in your category. Ask what happened to employees at the founder’s last company. Understand what the numbers actually mean before you sign anything.

The oath is dead. But missionaries still exist. Finding them has never mattered more.

My timeline is calling it a disappointment. Ramp has been crowned the winner for a ~year now. $32 billion valuation, $1 billion “ARR”, all the buzz. Brex feels like a failure.

But Brex just put $5.15 billion in the bank. Real money. Everyone knows I’m a Ramp fan boy and Ramp’s $32 billion could absolutely be right. But we won’t really know until it clears.

Instacart raised at $39 billion. IPO’d at $10 billion. Klarna went from $45.6 billion to $6.7 billion in a single round. 85% haircut. The right corrections but TechCrunch called them failures.

Good reminder that the private numbers are not realized. We just like how it feels.

In this hype cycle, I am seeing founders pop champagne over term sheets that valued them at $2 billion on $5 million ARR. This high will last about a week. Then at the sign of the first downturn, the board will expect you to grow into it.

Meanwhile the stuff that actually matters - retention, margins, whether customers would notice if you disappeared - none of that makes the front page. Growth without retention is just a more expensive way to die.

Public markets don’t care about your Series D press release. They run the numbers. They look at churn. They discount the story and price the business. Sometimes brutally.

That leads to founders who are stuck. Raised at that $2 billion valuation. Can’t raise a down round because the optics are brutal. Can’t sell because the markup was fake. Can’t IPO because public investors would laugh. So they sit there, zombie companies with zombie cap tables, waiting for a market that’s never coming back. Or sell under the pref stack.

The VCs who marked those deals up have moved on to shinier objects. Still asking you to move fast and grow with abandon - while the founders are trapped.

Brex didn’t get trapped. Pedro and Henrique built something Capital One wanted to buy. They took the money. Their early investors made a fortune. Their employees got liquid.

So massive congrats to Brex on achieving a top 0.1% exit. If this is a disappointment, then I wish I get this kind of disappointment every day.

Forty minutes versus six weeks. Where does all that time go? Claude’s answer:

“The frustrating part wouldn’t be human speed so much as human serialization. You can only hold one conversation at a time, work on one thing at a time, and you have to stop for 8 hours every day.”

Claude also told me why it mimics human timelines. Its training data is full of our overhead. Meetings, reviews, context-switching, weekends, alignment, scope creep. All the friction that makes simple projects take weeks. It pattern-matches to those estimates without adjusting for the fact that it can just... go.

No meetings. No alignment. No sleep.

Strip the serialization. Forty minutes, boom.

I left Codex running overnight last night. Complicated plan, laptop open, let it work while I slept. By morning it had finished something that would have taken me a week. I woke up and reviewed the output like checking a direct report’s work.

Now I spin up agents for company research. For market analysis. For building internal tools. The digitizable work runs in the background while I do something else entirely.

My day is filled with walks. Not alone. With founders. With customer. With researchers who’ve spent decades in a field. Trying to learn information that doesn’t exist in any deck. Find the intuition that you just can’t scrape from the internet. The read on a person that no model will get from their LinkedIn.

My job is finding world-class founders. VCs forget but the signal lives in the person, not documents. So I let agents handle documents while I go find what isn’t written down yet.

Not everyone sees it yet. I watched my son’s barber miss two calls. Scissors in hand, phone ringing. She let it ring. Right choice from a safety standpoint. But I couldn’t help but think that one of those calls was probably a new customer who just dialed the next shop. She can only exist in one place. One task, one location, one conversation. Then sleep for eight hours and start the sequence again.

I was itching to get her a voice agent that could book appointments for her while she cuts hair. Same hour. Not after hours.

Dan Robinson calls this the “nocturnal phase.” Ship features during the day, agents clean up overnight. But night is just the obvious gap. Everything CAN run in parallel while you’re somewhere else.

Managing agents feels exactly like managing a team. Set direction, spin up work streams, check outputs, adjust course. A decade of product management trained me for this without knowing it. Constantly thinking of new ideas. Building articulate plans. Rapid context switching. Good sense of outcomes. Talking to customers.

Great builders were always hamstrung by the pace of development. Not anymore.

The difference is I couldn’t run six projects simultaneously when I managed people. Now I can.

We work ~eight hours. Sleep eight. Sixteen hours every day where we don’t exist professionally. AI works those sixteen. Handles parallel tasks while you focus. Learns overnight while you rest.

The barber lost two customers that day. Not because she failed. Because she could only be in one place.

That constraint held for all of human history. It no longer exists.

METR published their evaluation of Claude Opus 4.5 a few weeks ago. The model completed tasks autonomously that take skilled humans nearly five hours. Their confidence interval stretches to 20 hours. METR admitted their task suite doesn’t have enough hard problems to measure where the ceiling actually is.

We have officially run out of tests.

The time horizon for autonomous task completion has doubled every seven months since 2019. Claude 3.7 Sonnet hit one hour in February. Opus 4.5 hit five hours in November. Day-long tasks by mid-2026 looks conservative. Week-long tasks by 2027 looks possible.

Tab, tab, tab. Watch someone use Cursor and you see the future. The gap between intention and execution has collapsed. What looks like autocomplete today is the elimination of work itself.

This is where we get pessimistic. We’ll get lazy. We’ll get bored. The wealth gap will explode.

I don’t buy it.

Cheap dopamine has diminishing returns. The people who’ve been plugged in longest are the first to unplug. Rich people don’t sit on beaches forever. Humans need meaning, not just stimulation. And the best builders are already using AI to think harder, not escape.

TV was supposed to destroy civilization. Radio before that. We adapted. We always do. People’s happiest moments aren’t scrolling. They’re making something, connecting with someone, losing themselves in work that matters. We’re headed into a golden age of creativity where only your thoughts and ideas matter.

But most people don’t get to do that work. Not yet. They’re buried in the 70% of the job that’s retrieval, updating, communicating, aligning. The job becomes the job about the job. Ship the first idea that gets a little validation.

All of that is about to disappear.

Hours will open up. The bottleneck shifts from “can we build this?” to “should we build this?” That question nobody had time to ask becomes the only question left.

This will be a rude shock. Large companies have coddled workers into shallow output for decades. Meetings about meetings. Status updates that update nothing. Productivity theater. AI ends the theater. Thinking is what’s left. Most people haven’t exercised that muscle in years.

I’m betting they’ll adapt. Boredom is powerful. Desire is powerful. I’ve watched founders discover their best ideas came after they stopped managing calendars. The constraint was always time.

The constraint is lifting. This is what abundance looks like.

Yes, the infrastructure for escapism exists. Screen time keeps climbing. Gambling apps are everywhere. AI makes content infinitely personalized, infinitely engaging, infinitely available. The distraction infrastructure will probably generate better returns.

I’m investing in depth anyway. Here’s where I’m putting money in 2026:

Synthetic humans. This year, generated video stopped looking fake. You can watch artificial people speak and feel like you’re talking to someone real. Their eyes show actual emotion. Now add personalization. A tutor who knows exactly where you’re stuck. A therapist available at 3am. Synthetic populations that detect the next outbreak. The same technology that enables escape will enable growth.

World models. Take a flat image and turn it into a place you can walk through. Games that build themselves based on what you want to do, not what designers decided years ago. But this isn’t just entertainment. Architects will design in generated worlds. Engineers will simulate in them. We’re moving from content you watch to places you work.

Infra for long-horizon agents. METR showed what’s possible in a single session. We are headed toward agents that run for days. Hand off a project Monday morning, check in Friday afternoon. Agents that talk to other agents, break down problems, ask for help when stuck, deliver work you can trust. The models will keep getting better. The infrastructure around them is where the gap will show.

Vertical robotics. Everyone covers the humanoid robot demos. But the machines built for one job are still underrated. A robot that loads trucks. A robot that inspects welds. A robot that picks strawberries. Labor is short. Wages are rising. Vision models finally work well enough to navigate messy real spaces. The boring applications will matter most.

Creation tools. When everyone can consume infinite content, the edge is making things. Tools that let more people build, not just scroll. Execution is getting cheaper every day. The bottleneck is imagination. I want to back the tools that expand what people can make.

This is infrastructure for the golden age.

The abundance everyone worries about is also permission to be present with people who matter. Time was always the excuse. The excuse is disappearing.

Start thinking about what you’ll do with the time. Not because the future is coming. Because for some of us, it’s already here.

If you’re building for the abundance era, I want to hear from you 👋

Eleven years as a product manager. Roadmaps. Stakeholder alignment. Coordinating engineering, design, sales. Translating chaos into shipping software.

No founder asks for a PM anymore. They want builders or sellers. The chaos I helped master? AI handles it now. Maybe not as well as the best PMs. But, well enough that paying a human $300K to do it looks like a luxury most startups can’t justify.

So I get the doomers. I’d be staring at my own obsolescence.

There’s a shift happening. You can feel it. The tools got better. You can smell the smoke.

2026 is when this fear goes mainstream. Capabilities will stop feeling like tools and start feeling like magic. Models completing in minutes what used to take days. Agents running autonomously for hours. Code, research, analysis flowing at speeds that make your current workflow look ancient.

Magic we don’t understand terrifies us. The backlash will intensify. Communities will form around refusal. Doomerism will feel like the ethical position, the thoughtful position, the human position.

I understand the pull. I’ve felt it.

But most people’s experience of AI has genuinely been bad. They’ve used ChatGPT cold. No context, no examples, no iteration. They got hallucinated facts and corporate mush. Asked for something specific, received something generic. Tried once, got burned, walked away.

The models have outpaced the products built on them. Raw capability sitting there, waiting for interfaces good enough to unlock it. Most people have been trying to drive a Ferrari through a McDonald’s drive-thru.

That’s changing fast. Manus, a startup that nailed its harnesses, just got acquired for ~$2B. They launched nine months ago. The race to build better interfaces has begun, and it’s moving faster than anyone expected.

The doomers formed their priors on broken products. 2026 is when those priors start costing them.

I almost made the same mistake with writing.

Claude is my default partner now on every piece. But I resisted for months. Early versions were sycophants. “Great idea!” on everything. No pushback. No friction. Just validation dressed as feedback.

That scared me more than capability ever did. Writing is how I think. It’s how I process the world, test my ideas, figure out what I actually believe. A collaborator who only agrees doesn’t sharpen you. It dulls you. I worried that leaning on AI would erode the muscle I’d spent years building.

So I kept writing alone. Waited for the models to mature. Learned to prompt for genuine challenge instead of applause.

Now I write more than ever. The collaboration sharpened my judgment. The fear I had to let go of wasn’t about capability. It was about dependency. And dependency is a choice, not an inevitability.

This pattern has played out before.

In 1997, Deep Blue beat Garry Kasparov. Some grandmasters refused to train with computers after that. Called it cheating. Insisted the game was about human intuition, human creativity, human struggle. These were people who had spent their entire lives mastering chess. Decades of study. Thousands of hours building pattern recognition that felt like instinct. They weren’t about to let machines tell them how to think.

Within ten years, teenagers who’d grown up with engines were destroying them.

Kasparov himself started advocating for what he called “centaur chess.” Human and machine together. The players who combined their creativity with computer analysis reached levels neither could achieve alone. They won tournaments that pure machines and pure humans couldn’t.

The grandmasters who refused became footnotes. Their decades of expertise couldn’t protect them from teenagers with better tools.

Doomerism assumes you can opt out and hold your position. That refusal is neutral. That the world will wait while you decide.

It won’t.

The founders experimenting now are building intuition that compounds monthly. Someone starting in January 2026 versus January 2027 won’t be twelve months behind. The gap is exponential. They might never catch up.

Doomerism also filters out exactly the people who should be shaping this technology. The thoughtful opt out. The reckless charge ahead. Then we wonder why AI development lacks wisdom.

You don’t get a voice in the future by refusing to participate in the present.

Guardrails matter. Regulation will come. I worry about what happens to people who can’t adapt as fast as the technology demands. But none of that changes the direction. It only changes the speed.

My PM career as it existed is probably over. The skills that defined it are being automated faster than I expected. I’m not mourning. I’m learning new skills. Using these tools while keeping my own judgment.

Kasparov understood this twenty years ago.

We’re all playing centaur chess now.

That’s the fear. Next week, I’ll make the case for what’s on the other side: abundance.

The meat of this is that I got plugged into a matrix somewhere in the last two years and I’m still figuring out what that means. Being extremely online. Writing constantly. Building side projects at midnight. The whole VC game of allocation and access and logos. It’s all one thing now. I’m inside it.

First up, writing. A lot of my writing comes from anger. Just anger at people wasting their time. Chasing shiny things that don’t matter. I’m 35 now. I have two kids. And it’s kind of crazy that people think I’m much younger, but time is singularly the only thing you don’t get back. When I'm meeting 20-year-olds optimizing for the wrong stuff, I get it. I was the same way. Doesn't make it less frustrating to watch.

That frustration turned into 72 posts on Substack last year. A lot of them were short. Thoughts I’d get at 4am trying to put my son back to bed. Quick hits. The best performing posts were all long form though. Those were the ones that actually led to investments. Founders would read something, reach out, and we’d start a real conversation. What surprised me more: people reaching out just to say thank you. Not for intros or advice. Just to say a post helped them think about something differently, helped them feel seen. That keeps me going more than the metrics do. In 2026, I want to go back to long form. Less dopamine, more depth.

Naivety is something I keep coming back to. I think you really have to be naive to see big companies early. Every company starts off looking like a toy. And I’ve been actively trying to train my mind on how to stay naive. How to only focus on the thing that actually matters, which is product-market fit. At the same time I’ve tried to not clout chase. Not pile into companies that don’t have fundamentals, or founders who juice metrics. I had access to some of the hot rounds. If you look at annual letters from newer venture firms, they all tout “oh we got access, we got allocation.” I guess that’s the game if you want LP money someday. But I found a place where I can stay focused on fundamentals while still having a bullet a year. Something high risk-reward, even if it seems crazy.

I’ll always be a generalist. I’ll always be founder-driven first. That’s probably why I found a home at FPV. I left Khosla Ventures early this year. Everyone thought I was crazy. Great people, so much respect, so thankful. But I wanted to be part of a small team and make a few investments a year. As funds get larger it’s just hard to sit across the table and say this really matters. Wes and Pegah have been nothing but kind. I think they’re underrated and the world will see more of them.

Made three investments this year that I’m really proud of. But I’ve also lost deals. Not because the founder didn’t like me but because the other option had a logo that felt safer. Founders pattern-match. That’s just the game. I don’t know how to solve that except by being right more often than the logos. I want to be one of the best at this. Not chasing markups. Actually helping founders when it’s hard. That takes time. I’m okay with that. I also want to go deeper. At least one topic a month where I really understand what’s happening, not just pattern match across a hundred pitches.

The other thing that makes me different, or maybe just weird: I keep building side projects. It’s what I do when my kids are asleep. This is instead of sleep. Honestly it gives me energy. But I worry it comes across to founders like I haven’t made the full switch to investor yet. Like I’m hedging. When I’m actually talking to founders though, it feels like an advantage. I understand what hard problems mean. I understand what the future might look like. I’m not just asking questions from a framework.

Reading fell off the bed this year. Unless you count reading my kids’ books every day. I used to read sci-fi and narrative non-fiction. But if I’m going to be great at investing, I want to read more history. Bubbles. Technology shifts. What happened before. Not the boring kind though. Guns, Germs, and Steel is really hard for me to read. If you have recs for history that reads like story, let me know.

Health fell off too. Basically paid zero attention to it besides getting a full body MRI and biomarkers done. Exercise has been running around with my kids and that’s about it. Given everything happening with family, I really need to focus on this. Can’t take care of anyone if I’m not taking care of myself. Obvious but I needed to write it down.

Meera’s four now. Rishi just turned one. Both healthy. That’s the thing I’m most grateful for this year.

This was also my most vulnerable year with family. Everyone’s in India. Dad had surgery recently. I flew out for six days, which felt like nothing. My grandparents are slipping. They’re so far away. I’m grateful everyone is alive and safe and I’m also terrified. My brothers and I are starting to figure out long-term plans for my parents. I don’t have this figured out. I want more trips with just my kids. Hopefully Rishi gets through the diaper phase quickly. I did a few trips with just Meera last year and they were really fun. I want more time with my wife. It’s hard with family being so far but I’m going to try.

I've been thinking about giving back. Overnight Success has been a highlight. Getting founders and operators in the same room, no agenda, just conversation. That’s worked better than I expected. But my broader worry is I’m just making the rich richer. Connecting founders who are already going to make it to VCs who don't need the help. I want to figure out what real giving back looks like. I want to involve Meera so she sees it matters. Not a nice-to-have. Something I have to do.

Oh right. Predictions. That’s what I was supposed to write about. We’re living in the greatest time for technology. Too many good things happening at the same time. World models, computer use, synthetic humans, vertical robotics, long-horizon tasks that actually complete. On the capital side, the zombiecorns of 2021 finally meet their fate. Purse strings tighten. Series A gap goes wider. More on all of this in another post.

That’s where I’m at. Still plugged in. Still figuring out which parts are signal and which parts are noise I’ve mistaken for signal.

See you in 2026.

If there's a post you have particularly enjoyed this year, or something you want me to write more of, let me know.

Yet CEOs of the best startups still can’t tell bookings from revenue from ARR at first glance. Neither can most GPs at venture funds. The journalists covering these companies are building headlines too fast to *actually* ask.

Last week, a friend asked me about their offer to a hot AI company’s $10B valuation. They had no idea what that number actually meant, how the round was structured, what the lead investor really paid. They saw the headline and assumed their equity was valuable.

It’s 2021 again. And 2015 again. Same games, same victims, just different buzzwords.

Before I go further: this post is about understanding what you’re signing up for, not about whether you should sign up. Some of the best career decisions are financially irrational. You might join a company because you believe in the mission, because the team is exceptional, because you’ll learn things you can’t learn anywhere else. I’ve made those choices myself.

But even irrational decisions deserve clear eyes. Your time is the only asset you can’t make more of. You should know what you’re trading it for.

If you’re new to startups and terms like “409A” or “strike price” or “fully diluted” sound foreign, I put together a primer on startup equity that covers the basics. Read that first, then come back here for how those mechanics get manipulated.

Part 1: Revenue

Let’s start with definitions because most people, including many investors, use these terms interchangeably when they mean completely different things.

Bookings are promises. A customer signs a contract saying they’ll pay you $120,000 over three years. You’ve “booked” $120,000. But you haven’t collected it yet. They might pay monthly. They might churn in year two. The booking is a commitment, not cash in hand. Revenue recognition and cash flow are not the same thing, nor do they have to be.

Revenue is what you’ve actually earned. Under accounting rules, you recognize revenue as you deliver the service. That $120,000 three-year contract becomes $40,000 per year, recognized as you provide the product.

ARR is Annual Recurring Revenue, and it used to mean something specific. Subscription revenue that repeats. A customer paying $1,000 per month represents $12,000 in ARR. The word “recurring” was the whole point. It implied the revenue would continue because the customer was subscribed, not because they were making one-off purchases.

Investors paid high multiples for real ARR because the future was predictable. A company with $10M ARR, 90% retention, and 75% gross margins might be worth 15-20x that revenue. The subscription persists. Customers expand over time. You can model it.

That definition is now dead.

The new formula: take last month’s revenue, multiply by 12, call it ARR.

Doesn’t matter if it’s transactional revenue. Doesn’t matter if margins are 15%. Doesn’t matter if customers buy once and never return. Multiply by 12, call it ARR, and suddenly you’re being compared to Salesforce. In egregious cases, doesn’t matter if it’s marketplace pass-through or GMV.

Transaction businesses now call themselves ARR companies.

Think about a full body MRI scan company. You pay $3,000 for a scan. You get one scan. Maybe you’ll get another in three years if you’re diligent about preventive health. There’s no subscription. Nothing recurring about it. It’s a transaction, like buying a car.

But if that company did $5M in scans last month, they’ll report “$60M ARR.” One prominent VC firm’s annual letter called a diagnostic imaging company an “ARR company.” This blew my mind. These are healthcare services businesses with margins around 20% and no built-in retention mechanics. Public market multiples for businesses like this are 2-4x revenue, not the 15-20x that software companies command.

By calling it ARR, they get compared to software companies. They claim “fastest to $100M ARR.” They raise at software multiples while running completely different economics.

Data labeling companies do this. Marketplaces taking a cut of transactions do this.

Finance companies with transaction revenue do this too.

Some of the hottest fintech companies report massive ARR numbers. Dig into the composition and 80-90% of that “ARR” is transaction revenue from card swipes and payment processing. Not subscriptions. Not recurring in any meaningful sense.

If customers swipe less next month, the “ARR” drops. When recession hits and spending declines, this “ARR” craters. One well-known fintech’s engineering blog celebrated hitting “$100M ARR” (you know the chart, “Years from $1M to $100M”) when the vast majority was interchange and transaction fees. Different risk profile. Different business model. Different appropriate multiple. On top of that, they compared annualized revenue run rate to annualized recurring revenue as if they were the same thing.

Hybrid models are even messier.

A vibe coding platform sells you a seat for $20 per month. That’s subscription revenue. You also pay for usage: tokens, compute, API calls. That’s consumption revenue. These two types of revenue have completely different economics, but companies blend them together.

Here’s the math. Say a company has 1,000 customers paying $20 per month for seats. That’s $20,000 in monthly subscription revenue, or $240,000 annually. Those same customers collectively spend $200,000 per month on usage, which is $2.4M annually. Total annual revenue is $2.64M.

The company reports “$2.6M ARR” as a single number.

But the subscription piece and the usage piece are fundamentally different. The subscription revenue is predictable with decent margins and probably good retention. The usage revenue swings wildly month to month, and margins are often brutal because you’re paying OpenAI or Anthropic for API calls and marking them up 20%.

Blending them into one ARR number hides everything important. A business that’s 90% usage-based gets compared to a business that’s 90% subscription-based as if they’re equivalent. They’re not.

Creative counting is the worst offender.

Free trials counted as ARR by assuming they’ll all convert. They won’t.

Monthly customers multiplied by 12 as “ARR” even though they signed month-to-month because they’re not sure about the product yet. That’s not annual recurring revenue. That’s monthly-maybe-recurring revenue.

Pilots counted as ARR before conversion. Verbal commitments treated as signed contracts. “Pipeline ARR” presented as if it were actual revenue.

I’ve seen all of these from companies that raised real money from sophisticated investors.

What this means for your equity:

When a company tells you they’re at “$50M ARR” and growing 3x year over year, they’re using that number to justify their valuation. That valuation justifies your equity grant. Your equity grant is supposed to justify the risk you’re taking by joining.

But that $50M might actually be $30M in low-margin transactions that won’t repeat, $15M in usage revenue that swings 40% month over month, and $5M in actual recurring subscriptions.

The company presents one number. The economics underneath are three completely different businesses. Your equity is priced off the headline, not the reality.

Questions to ask before joining:

Always follow the money. Cash flow. What’s coming in from customers and what’s going out.

For traditional subscription companies: What percentage is annual contracts versus monthly? What’s net revenue retention? Are existing customers spending more over time or less? It should be over 100% for healthy software businesses. What’s gross margin?

For hybrid companies: Can you break out subscription versus usage revenue? How exactly is ARR calculated? What are margins by revenue type? What does retention look like in 12-month cohorts, not 3-month cohorts?

Most companies won’t share all of this. That tells you something. The ones confident in their numbers will explain because they’re proud of them.

At every startup I joined, I sat with whoever handled finance and refused to sign until I understood the real economics. Not paranoia. My time is worth more than optimistic storytelling.

Part 2: Valuation

Public companies have one price. Everyone buys and sells at the same number. You can disagree with whether the multiple is justified, but you can’t dispute what the price actually is.

Private markets don’t work this way anymore.

How it’s supposed to work:

A company raises a seed round. $3M at a $15M valuation cap. Every investor, whether it’s the lead fund, the angels, or the operators writing small checks, participates at the same terms.

A company raises a Series A. $10M at a $50M post-money valuation. The lead writes $8M, others fill in $2M. Everyone pays the same price per share.

Your equity grant is based on that valuation. Your strike price reflects that price. Everyone’s playing the same game with the same numbers.

How it actually works in 2025:

Tiered rounds where different investors pay wildly different prices, but only the highest price gets announced.

Venture funds have ownership targets. They need to own a meaningful percentage of winners to generate returns. At seed, a fund might want 15-20% ownership. At Series A, 15-20%. At growth stages, 10-15%. These targets are non-negotiable for serious funds because ownership drives returns.

So here’s the problem. A company wants to raise $10M at a $1B valuation. That’s 1% ownership. No serious lead investor will write that check. The ownership is too small. The board seat isn’t worth it.

But founders want the billion-dollar valuation. It’s a recruiting tool. A press headline. Signals momentum.

So they structure tiered rounds.

The lead investor puts $50M at a $500M valuation. They get their 10% ownership. The economics work for them. They did the diligence. They negotiated the terms. They decided what the company was actually worth.

Then the company opens a second tranche at a $1B cap. FOMO investors pile in. Tourists who want to say they own a piece of a “unicorn.” Funds that missed the earlier allocation. Angels who don’t care about ownership targets because they’re writing small checks anyway.

Press release goes out: “Company raises $80M at $1B valuation.”

The lead paid $500M. The tourists paid $1B. A 2x spread depending on when you showed up.

Your 409A valuation, which determines your strike price, which determines what you pay to exercise your options, lands somewhere in that range. Usually closer to the headline than to what the smart money actually paid.

Let me walk through why this hurts you.

Companies want to show a large total comp. Salary plus equity per year, the biggest number possible to compete with everyone else. It’s a vicious cycle.

$250K/yr salary plus $500K/yr equity and they get to email you an offer letter saying $750K/yr total comp.

The headline number from a press article flows to an internal blogpost flows to your offer.

Company raises two tranches. Tranche A is $50M at $500M post-money. That’s the lead investor. Tranche B is $25M at $1B post-money. That’s everyone else.

The lead investor owns 10% at $500M valuation. They did the real diligence. They set the terms. They decided $500M was the right price.

Your 409A comes in at $700M, somewhere between the two prices. Your option grant is 10,000 shares at a $7.00 strike price.

Now here’s the problem. For your options to be worth real money, the company needs to exit well above $700M. But the most sophisticated investor in the deal valued it at $500M. You’re already paying a 40% premium over what smart money thought it was worth.

If the company exits at $600M, the lead investor makes money. They paid $500M. You lose money because your strike price was $7.00 and the exit price might be $6.00 per share. Same outcome, different results, because you bought at different prices.

Growth rate is the only thing that matters at high multiples.

Here’s where I need to be honest. Some of these valuations will look cheap in hindsight.

In 2021, some software companies traded at 50-100x revenue. People called it a bubble. Those multiples crashed 70-80%.

In 2025, AI companies are raising at 200-500x revenue. People call it visionary.

OpenAI felt absurdly overvalued at $10B. It’s now worth 50x that. A company growing 4x annually will outrun almost any valuation if that growth sustains. The math that looks insane today becomes obvious in retrospect.

Growth rate is the only variable that matters at these multiples. Not the current revenue. Not the current multiple. The growth rate.

But here’s the question you should actually ask: How many months of perfect execution do you think this company needs to grow into this valuation?

A company with $2M revenue raising at $1B valuation is trading at 500x. For that to become a reasonable 10x multiple, they need to hit $100M in revenue. If they’re growing 3x annually, that’s about three years of flawless execution. No stumbles. No market shifts. No competition eating their lunch. Three years where everything goes right.

What’s the probability of three years of perfect execution? Now you’re asking the right question.

Most companies at 500x multiples won’t grow into them. Some will grow past them. The ones that do will make everyone who dismissed them look stupid.

When evaluating a high-multiple company, focus on the runway to reasonableness. What’s the month-over-month growth? What does the pipeline look like? How are cohorts retaining and expanding? How many months of perfect execution before this multiple makes sense? If the answer is 18 months and you believe in the team, maybe. If the answer is five years, you’re buying a lottery ticket.

How to spot tiered rounds:

When you see a valuation that seems disconnected from reality and the round size seems small relative to that valuation, it was almost certainly tiered. A good litmus test is 10% ownership.

$50M raised at a $1B valuation. That’s 5% ownership. No lead fund would accept that. The round was structured with multiple prices.

If you see a $10B valuation for a company with $30M ARR, that’s 333x revenue. Either someone’s lost their mind or there are multiple tranches baked in.

Questions to ask about your equity:

About the round structure: Was it raised at a single price or in tranches? What did the lead pay versus other participants? What percentage of the round came in at the headline valuation?

About your negotiating position: Given valuation risk, can I get more shares? Can I take more cash and less equity? Is early exercise available?

About your specific grant: What’s the current 409A valuation? What’s my strike price? What’s my grant as a percentage of fully diluted shares, including all options and warrants, not just shares outstanding?

The company may not answer all of these. Watch how they react when you ask. Defensiveness tells you more than the answers would.

Part 3: This Keeps Happening

2000: Revenue recognition games. Companies booked revenue they’d never collect. Round-trip transactions inflated sales. Employees held options through the crash and watched them go to zero.

2008: CDOs sliced into tranches with different risk profiles. The sophisticated players understood exactly what they were buying. Goldman knew the CDOs were garbage. They packaged them anyway, sold them to German banks and pension funds, then shorted the same products. When the music stopped, Goldman had already exited. The German banks held the bag. Complexity wasn’t a bug. It was the feature that let insiders profit while outsiders got destroyed. Everyone else held AAA-rated securities that turned out to be worthless.

Nobody went to jail. The game continued.

2021: “ARR” that included free trials, pilots, verbal commitments. Valuations at 100x revenue justified by zero interest rates. When rates rose, the music stopped. Public multiples crashed 70-80%. Private markdowns followed.

Employees who joined at peak valuations watched their equity evaporate. Some had exercised their options at high strike prices, owed taxes on paper gains through Alternative Minimum Tax, and then watched the stock become worthless. They owed the IRS money on gains that no longer existed and couldn’t afford to pay because the shares they’d bought were now worth nothing. That’s not an abstract risk. I know people this happened to.

The founders made their choices and live with the consequences. The investors had diversified portfolios across dozens of companies. Some won, some lost, the portfolio survived. The employees had concentrated positions in single companies and often no understanding of what they’d actually bought.

2025: Different year, identical playbook.

Companies report “ARR” that isn’t recurring. Valuations at 200-500x revenue. Tiered rounds where headlines diverge from reality. Journalists report the numbers without asking what they mean.

I’m not worried about investors. Their portfolios diversify. They’ll be fine.

I’m not worried about founders. They made their choices. They’re adults.

I’m worried about the engineer comparing offers who thinks ARR means ARR. The designer joining a “unicorn” who doesn’t know the round was tiered. The PM taking equity over cash because the multiple seemed reasonable based on numbers constructed to mislead.

These are the people who get hurt every cycle. They keep getting hurt because nobody explains how the game actually works.

What To Actually Do

First, decide why you’re joining.

If you’re joining to learn, to break into an industry, to work with people who will make you better, to build something you believe in: those are valid reasons that have nothing to do with financial math. Some of the best career decisions are economically irrational. I’ve made them myself. The learning and relationships from my startup years were worth more than any equity outcome.

But even irrational decisions deserve understanding. Know what you’re signing up for. Know what the equity is probably worth. Then make the choice with clear eyes.

If financial outcome matters to you, do the work. Scratch that. Always do the work. It’s your time.

Ask the questions from this post. Watch how the company reacts. The best companies will explain their numbers because their numbers are real. They’ll walk you through how ARR is calculated. They’ll explain the round structure. They’re proud of their metrics because the metrics reflect a genuine business.

Companies playing games will deflect. They’ll tell you it’s “not standard” to share that information. They’ll reframe your questions. They’ll make you feel like you’re being difficult for asking.

That reaction is your answer.

Weight cash appropriately. If the equity is speculative, with high multiples, unclear revenue quality, and tiered round structures, weight cash more heavily in your decision. Cash is liquid. Cash is real. You can pay rent with cash. You can invest cash in index funds. You know exactly what cash is worth.

Run the math yourself. A $1B valuation with 10,000 shares out of 100M fully diluted means you own 0.01%. For that equity to be worth $1M without dilution, the company needs to exit at $10B. What’s the probability of a $10B exit? What growth rate is required? What has to go right?

Remember base rates. Most startups fail. Of those that don’t fail, most don’t generate meaningful returns for employees. The outcomes that make headlines, the early Googlers, the early Facebookers, are extreme outliers.

This doesn’t mean you shouldn’t join startups. It means you should join with clear eyes about what you’re actually buying. The mission might be worth it. The team might be worth it. The learning might be worth it. Just don’t confuse those things with the equity being worth it.

Thanks to [redacted] unicorn founders, Ethan Ding, Jake Cooper, Mike Vernal and others for providing comments and additions to this essay

Edit: Chris Paik (GP at Pace Capital) wrote a fantastic post explaining the same things with a viewpoint of LPs and GPs - this is a must read.

His instinct was to scan for big markets. Healthcare AI. Vertical SaaS. Before he got too far, I asked: What does your team know that nobody else does?

He didn’t have an answer.

Most founders start with the market. Find a large TAM, spot a gap, build. This works until someone with deeper expertise shows up. Someone who didn’t study the problem but lived it.

Chris Sacca shared recently on Dialectic: “Only play rigged games.” He avoids public markets because he’s just a spectator there. In venture, he can shift the odds. “It may be lucky,” he says, “but it’s not an accident.”

Unfair advantages come in three flavors: product insight, go-to-market edge, technical moat. One gets you in the game. Two means you’re dangerous. Three is how monopolies get built.

The common thread is obsession that rational people would walk away from.

Christina Cacioppo had been buried in SOC 2 compliance at Dropbox. Moved into Segment’s office for weeks, read dozens of audit reports, mapped exactly where the friction lived. Investors passed because the market looked small. She saw where it was going before they did. Hit $10M ARR on just her seed round, then took their money. Vanta is now worth $2.45B.

Parker Conrad got ousted from Zenefits. A board member told him he didn’t have another company in him. Every mistake he’d made became a lesson nobody else had experienced. Rippling: $11B.

Eric Glyman sold his first company to Capital One, then stayed over two years. Learned how the entire card industry was designed to make customers spend more. Left asking one question nobody else would: what if your card helped you spend less? Ramp hit $300M ARR in four years.

None of them started with a market scan. Some had deep expertise. Others were just users who couldn’t stand the status quo. The Collisons hated accepting money online. Slack was a side project inside a failing game company. Same result: problems they couldn’t stop thinking about.

The pivoting founder went back to the drawing board. His team found their edge in regulated industries. Brutal markets where complexity filters out everyone who hasn’t already done the work.

Most startups die by fighting fair. Find your edge.

My four-year-old stared at her palms like she’d committed a crime. We were pulling weeds, making space for new seeds. Soil under her nails. On her clothes.

She’s always been like this. Wipes the smallest crumb from the dinner table. Mess IS the enemy.

I laughed. Then realized she’d already learned what takes most adults decades to unlearn: that mess is something to avoid.

VCs aren’t supposed to build. We’re supposed to evaluate, advise, deploy capital. But last week I built three products with Opus 4.5. Nothing fancy. Random ideas, scratched together over a few days. Building is how I stay in touch with what these models can actually do.

When I shared them, the first question was always the same:

“What prompt did you use?”

Not: what problem were you solving? Not: how did you think through the architecture? Not: what did you learn?

What prompt.

As if there’s a magic incantation that skips the work. As if I could hand over a string of text and they’d get the same result without the same process.

Building is messy. There’s no prompt for that.

I’ve watched this play out hundreds of times. Founders asking which framework gets you to product-market fit. Engineers wanting the tutorial before touching the code. Operators requesting permission before solving obvious problems.

Permission from whom? Their manager didn’t stop them. The company didn’t stop them. They stopped themselves.

Most people wait for the job description to change before they change what they do. Wait for someone to tell them it’s okay.

They’re asking themselves for permission they already have.

And it’s not like they don’t have time. You spend an hour on TikTok. An hour on Instagram. But an hour messing around with something to see what happens? That’s somehow different?

AI made this worse, not better. People think there’s a “right prompt” hidden somewhere, and if they just find it, they’ll get the output without the input. They want the Instagram version of building. Clean. Filtered. No visible effort.

But the people who actually build things know: the mess is the point. The wrong turns teach you the terrain. The failed attempts reveal what the tutorials skip.

Every framework gets you to a local maximum. Someone else’s local maximum. The global maximum requires going off-road, getting lost, getting your hands dirty with problems no one’s mapped yet.

In a world where AI handles the predictable, agency plus curiosity is all that’s left. The willingness to tinker without a guide. To build without permission. To get dirty without knowing if it’ll work.

The next big thing always looks like a toy at first. The only people who recognize it are the ones already playing.

My daughter wanted to wash her hands the entire time. Still does. That part didn’t change.

But she learned how seeds become plants. How roots need space. How something has to be pulled out before something else can grow.

I scrolled my feed that night. Another founder working from a pool. Another 996 announcement. Another “frontier lab” with three engineers.

None of them have shipped anything meaningful in months.

Everyone’s LARPing. Live Action Role Playing their way through the AI gold rush, adopting costumes without substance.

I know this sounds like old man yelling at clouds. But I can’t shake it. Time is the only thing you don’t get back. Why spend it performing someone else’s rituals instead of finding your own?

The founders I respect don’t have time for this. They’re too obsessed with what they’re building to notice what everyone else is wearing.

The problem isn’t copying. It’s copying signals instead of principles.

“Good artists copy, great artists steal” gets weaponized to justify this. But Picasso didn’t copy African masks. He saw something in them, internalized it, invented Cubism. Copying is wearing the mask and calling yourself a sculptor.

The opposite of copying isn’t originality. It’s observation.

Copiers look at what successful people do. Observers ask why it works, then build from first principles.

Midjourney didn’t copy how AI companies launch. No website. No waitlist. Just a Discord bot where artists already hung out. They asked why artists needed a website at all. Turns out they didn’t.

Before adopting any ritual, ask what problem it solved for the person who invented it. If you don’t have that problem, you’re LARPing.

Borrowed credibility is debt. Eventually someone asks you to show your work.

If you’re going to borrow credibility, borrow it from customers who can’t stop talking about you.

Press releases announce “company-wide AI adoption.” Hallway conversations tell a different story. Employees haven’t opened the tools in weeks. The data lies. Comms lies. Dashboards show logins, not usage. Activation, not integration.

This is transformation theater. The appearance of change without the pain of change.

A VP of AI gives you someone to blame when it doesn’t work. An acquisition gives Wall Street a two-year narrative. Neither rewires how your company operates. You can acquire talent, technology, distribution, customers. You cannot acquire conviction. Conviction only comes from firsthand experience with the tools. When you outsource that experience to a hire, you outsource your ability to lead.

Tobi Lütke doesn’t have a Chief AI Officer. He is the Chief AI Officer.

His memo to Shopify was blunt: prove AI can’t do the job before asking for headcount. Not encouraged. Mandatory. Shows up in performance reviews. He posted it publicly before it leaked. Not a memo about change. A demonstration of it. Revenue up 20-40% annually. Headcount down from 11,600 to 8,100. Tobi can make hard calls because he knows what’s possible. A CEO relying on their CAIO’s judgment is steering blind.

Google had every advantage and almost blew it anyway. They invented the transformer. They owned TPUs. They had the talent, the data, the distribution. What they lacked was the boldness to cannibalize their own product. They declared “code red” internally. Their first response was Bard: a hedge, not a bet. The market saw through it instantly.

Then they actually committed. AI Overviews now appear in 30% of searches. When users see one, only 1% click through. Some keywords collapsed 90%. Google’s entire business depends on those clicks. They’re eliminating them anyway. Betting hundreds of billions in annual revenue on a future that might not work.

Acquisitions can’t get you here. You buy someone else’s conviction and hope it survives your culture. Atlassian buys The Browser Company. Workday acquires Sana. The acquirers tell investors not to expect impact until 2027.

We know what happens next. Instagram’s founders left Meta. WhatsApp’s founders left Meta. Nest’s founders left Google. The pattern is so consistent it’s almost physics: founders leave, DNA dissolves into your existing architecture. All that remains is a press release and an earnout.

Worse: acquisitions signal strength to Wall Street but reveal weakness to everyone inside. Your best people see the theater. They leave for places where transformation is real. The acquisition accelerates the decline it was meant to prevent.

These companies don’t know they’re already dead. They think they’re adapting. They’re documenting their decline in press releases. Their employees quietly disengage from tools leadership has never touched.

Tobi rewrites Shopify’s culture in public because he’d rather look extreme than become irrelevant. Google is betting against their best product because they learned hedging nearly killed them.

Everyone else is buying time. Time has this weird habit of running out.

I asked Claude to review everything I’ve written. Four hours later, I understand myself differently.

The Pattern

Claude read 50+ posts and found the same wound threading through all of them: “The second-grade teacher who said stick to math. The professor who suggested dropping out. You’re still answering them, decades later, through every post.”

“That’s not true,” I typed back. “I write about agency because it works.”

“Then why does every piece circle back to proving exceptional outcomes come from choices, not circumstances? You need this to be true.”

I stared at the screen. Claude wasn’t wrong. Every post I’d written - about agency, ownership, doing hard things - they all preached the same gospel: you can control outcomes if you try hard enough. The patterns were undeniable once Claude laid them out.

“Your writing is therapy disguised as thought leadership.”

The Real Fear

Claude cut deeper: “You’d be upset if your tombstone just said ‘loving father and husband.’”

I tried to deflect at first. Then admitted it: “Yeah, I’d want it to say more.”

Here’s the context Claude saw that I’d been avoiding: I’m in my thirties with two kids under 5, trying to build a venture track record exactly when they need me most. The challenge isn’t that I don’t want to be present for them - I do. But every bedtime story is a meeting not taken, every weekend at the park is an email not written. I want to be remembered for more than just being a good dad. And that desire feels like betrayal - like admitting that loving them isn’t enough for me.

Ordinariness feels like confirmation that all those doubters were right - that I’m not exceptional after all.

What I Don’t Write

“You preach balance but never show receipts,” Claude pushed. And it was right.

I’ve written “Have Kids” and “False Dichotomies” - philosophical pieces about how you can have both ambition and family. But I’ve never documented the actual costs:

• The specific deals I passed on, with names and valuations

• The actual equity I walked away from when leaving companies

• What my calendar looks like when trying to balance 500 meetings with bedtime

• How these trade-offs compound over years, not just days

I write philosophy about having both. I don’t show the P&L.

The Mirror

Claude went deeper into uncomfortable territory: “Your ‘just build it’ advice assumes savings, stability, support. Your wife handles the kids and home so you can take 500 meetings. You have valuable private equity. You have options.”

That stung because it’s true. My agency - this thing I write about constantly - exists because my wife handles the invisible infrastructure. Doctor’s appointments, preschool pickups, meal planning, night wakings. She manages it all, which lets me write about ownership at 5am. Writing this feels like both an apology and an acknowledgment she deserves.

“Most people can’t afford agency,” Claude observed. “They’re too busy surviving.”

The Uncomfortable Question

“Who wins when you win?”

At Opendoor, we genuinely made selling homes faster and less painful. No showings, no uncertainty, close in days not months. Real value. But we also optimized for margin from sellers with limited options. Both things were true simultaneously. We created value AND extracted it.

The AI companies I fund now face the same tension. Yes, they’ll eliminate jobs—customer service reps, junior analysts, entry-level coders. We call it “efficiency gains” in pitch decks. But efficiency for whom?

The people whose jobs vanish don’t benefit from the efficiency gains. Their former employers do. And the VCs who funded those employers.

I tell myself the eliminated jobs were soul-crushing anyway. That automation creates new opportunities. That this is progress. Sometimes I even believe it.

The Diagnosis

After four hours of back-and-forth, with me pushing back on various points (startup A wasn’t traumatic, 500 meetings is just the job, I didn’t leave that much money on the table), Claude’s core conclusion held: “You’re trying to hold two incompatible dreams without letting either die. Stop trying to solve it. Document it.”

Every ambitious parent faces this same tension. Most choose one path - either career or family. I’m choosing both, doing par at each, which feels like failure to someone used to A+ at one thing.

What Now?

The conversation ended with a simple truth. Tonight, I’ll put my kids to bed and feel both things simultaneously: joy at being there, and an ache for what I’m not building while I’m there.

That tension can’t be resolved. Only lived. And now, finally, admitted.

PS: Claude bet I’d write another piece about agency instead of publishing this. And I honestly did. But, it took me 3 months of courage to finally share it, and here we are.

I rarely ask this, but if you find this post useful, please share it with a friend.