The Next Epoch of Venture.

Every investment is a bet on a picture of the world. You look at a company, but you price a future: what will be scarce, what will be abundant, who will matter, and why. Most of the time the picture drifts slowly enough that nobody has to repaint it.

This is not most of the time.

Over the past few years, the cost of better intelligence itself — analysis, drafting, synthesis, translation, the everyday work of "knowing things" — began falling toward zero. Not the cost of wisdom. Not the cost of trust. The cost of processed information. And nearly every institution in the modern economy, from the law firm to the org chart to the venture fund, was built on the assumption that processed information into action is expensive.

That assumption is dying. Quietly, unevenly, and faster than the institutions built on it can admit.

This mainfesto is our attempt to face that squarely. It is not a forecast, and it is not a brand exercise. It is a chain of reasoning: what actually changed, what it does to companies, what it does to venture, and why we believe the next great investment firms will be built — deliberately, structurally, unsentimentally — on human capital.

Not as a slogan. As the asset.

Start with a question economists asked a century ago and most of us stopped asking: why do companies exist at all?

The honest answer is friction. It was expensive to find the right person, verify their work, negotiate terms, coordinate the pieces. So we built organizations to swallow that friction — layers of managers, analysts, reviewers, and coordinators whose real job was never to make the product. Their job was to move information: translate it, check it, route it, summarize it upward and delegate it downward. The modern company is, in large part, a patch to compensate for the fact that knowing things used to be hard in a previous era of civilization.

The idea of a company was never a law of nature. It was a workaround for collaboratively turning expensive information into focused action.

So what happens when knowing things becomes nearly free?

Not the collapse of the company — the collapse of the patch. Work whose entire value was processing information faster than the next person loses its margin, inside companies and between them. The middle of the organization thins first, because the middle is mostly translation. What holds its value sits at the two ends: judgment, deciding what's worth doing and what's true, and craft at the edges, where things actually get made and customers actually get met.

But even "knowing things became free" undersells what's happening, and it's worth slowing down here, because this is the part almost everyone is still underestimating. Not all knowing is the same kind of knowing. There's knowing facts — what's true. There's knowing how — the steps to do something. And for all of human history, that's all our technology could hold. Books held facts. Databases held facts. Software held steps. But there are two deeper kinds of knowing that never left the human head: knowing what matters — the perspective to look at a messy situation and see which of a thousand details is the one that counts — and knowing by being in it — the understanding you can only earn by participating, by doing the work and being changed by the doing, the way a craftsman knows wood in a way no manual ever will. Companies are secretly built out of those two deep kinds. Judgment is knowing what matters. Craft is knowing by being in it. Everything else was always just paperwork in between.

Here's what's new — actually new, not faster-and-cheaper new. This technology is the first in history that doesn't stop at facts and steps. It can hold a point of view. It can weigh what matters in a situation it has never seen before. And it can take part in the work itself — not a reference you consult, but something closer to a participant in the room. Every previous technology knew about the work. This is the first one that can be in the work.

We should be careful with anyone who claims to know where this lands, ourselves included. Every structure inside the modern company — every role, every layer, every approval chain — was designed on the quiet assumption that only humans could hold participatory and perspectival knowledge. That assumption just broke, and we are at the very beginning of discovering what follows. When electricity arrived, factories first bolted motors onto their old steam-era layouts and saw modest gains; the transformation came a generation later, when someone finally redesigned the factory around what electricity made possible. That's where we are: the bolted-on phase of something much deeper. The companies being started today are the first ones that get to be designed, from the ground up, around a workforce that includes machines that can see and take part.

Even a small shift in how information is managed can make companies smaller and sharper. A team of eight with taste, technology and trust can now do what once required an entire building of people. Companies will shrink in headcount long before they shrink in ambition, and the org chart starts behaving less like a pyramid and more like a project: gather the right people around the problem, build, dissolve, gather again.

Let's give venture its due before we take it apart. For the last era of civilization, venture capital firms, angel groups, syndicates, etc were valuable infrastructure. Venture stood in that gap and moved money toward ideas the banks wouldn't touch. It worked. Much of the modern world got financed through exactly this machinery. But infrastructure is only valid as long as the scarcity it manages still exists — and the scarcity venture was built to manage is disappearing underneath it.

Because be honest about what the investing business — venture, angel, private equity, all of it — has actually run on this past decade. Access: the warm intro as the first filter, the brand-name firm as gravity, the best deals reserved for whoever was closest to the room. Networks: who you know functioning as diligence. Pattern matching: a partner's personal sample of a few hundred companies, projected onto the next ten thousand. Social proof: other investors' conviction treated as evidence, markups treated as validation, momentum treated as truth. And above all, signal-finding — thousands of pitches a year, screened at speed using shortcuts: the polish of the deck, the pedigree of the founder, the shape of the growth curve, who else was already in.

Strip it down and you see what the business really was. Capital was the commodity; anyone's money spends the same. The actual product was filtration — finding signal in a sea of noise.

Every one of those signals worked only because it was hard to fake. A crisp deck stood in for clarity of thought. Smooth answers stood in for deep understanding. A credential stood in for years of demonstrated work. The filter was calibrated to a world where producing convincing signal was expensive — and that world is gone. Any founder can now generate the perfect deck, the polished model, the fluent answers to every diligence question, in an afternoon. This isn't dishonesty; it's the new baseline. But it means a filter built to detect signal now mostly detects tool fluency. The sea of noise didn't get louder. It learned to look exactly like signal.

Now, the tempting move — the one half the industry is making right now — is to take the old process and automate it. Feed the machine our pattern matching, let it screen the decks, score the founders, rank the deals. Faster funnel, same logic. It's worth thinking carefully about what that actually produces, because the answer is worse than nothing.

Automate the old filter and you carve yesterday's shortcuts into a system that never questions them. A model trained on who got funded doesn't learn who deserves funding — it learns who used to get funded, and repeats that history at industrial scale, with the errors and blind spots baked in permanently. And since every firm's machine reads the same public signals, they all converge on the same companies: herding, at machine speed, priced accordingly. Meanwhile the founders' machines are generating precisely what the investors' machines are screening for, and you arrive somewhere genuinely absurd — an economy of machines impressing other machines while the actual company goes unexamined. Diligence theater, performed at light speed. And the outliers — the strange, unpatterned companies that venture exists to find, the ones the power law depends on — are exactly what pattern-automation is built to screen out.

AI pointed at yesterday's process doesn't modernize venture. It fossilizes it — the same mistakes, faster, forever.

So the way through is not the old filter with a bigger engine. When the signals can all be generated, the only thing left worth examining is what can't be: how a founder actually thinks when the plan breaks, whether a team actually tells itself the truth, how fast either one actually corrects. The rest of this manifesto is about that — what it looks like to underwrite the part that can't be faked, and to build machinery that examines it honestly. For some firms, when the old moat drains, the honest answer to "what was underneath?" will be nothing. This is what it takes for the answer to be something.

If you think this jump in intelligence is the equivalent to the creation of the computer or the internet, you are underestimating how much will change in the world when intelligence becomes as available as electricity.

Gutenberg could never have predicted the teleprompter.

Cycles are fluctuations within a stable structure — prices swing, seasons turn, the structure holds. What is happening now is the structure itself moving, and not along one axis but several at once: machine intelligence, energy systems, geopolitics, demographics, industrial policy, the plumbing of finance. When one system shifts, precedent is a guide. When so many markers shift simultaneously and interact, precedent becomes a comfort blanket with predictive value approaching zero.

Sit with that for a moment, because it quietly breaks the industry's favorite tool. The instinct, when the world gets uncertain, is to predict harder — more data, more conviction about what happens next. But when the ground itself is moving, the premium isn't on prediction at all. It's on orientation: holding a small number of durable beliefs about how systems, incentives, and people behave, and reasoning freshly from them as the terrain changes underfoot.

Forecasts expire. Principles compound.

Orientation redraws the map. Sectors — fintech, healthcare, climate — are filing cabinets from the old world, useful for organizing analysts and conference agendas. The defining companies of this era won't be found inside those drawers; they'll appear where systems collide — where energy meets compute, where demographics meet automation, where trust meets infrastructure. And orientation changes the toolkit too: sometimes a thesis wants equity, sometimes credit, sometimes something closer to infrastructure. The instrument should follow the idea. An investor whose identity is an asset class has confused the hammer for the house.

There's a pattern that shows up every time technology takes a leap, and once you see it you can't unsee it. Each shift makes one thing abundant, and in doing so, reveals what was scarce all along. Photography made images cheap — and made the photographer's eye precious. The calculator made arithmetic free — and made mathematical taste the differentiator.

Machine intelligence makes answers cheap. It makes questions critical.

Think about what that does to a company. When execution is abundant, the entire weight of the enterprise shifts onto knowing what to ask for — what's worth building, what's actually true, what matters most right now. And that is not an information skill. You can't download it. It lives in people: in their clarity, their character, their capacity to stay honest under pressure. It compounds or corrodes depending on how those people live and work together.

Now follow the logic one more step, because this is the hinge of the whole manifesto. If the scarce input to value creation is no longer information, and no longer execution, then it's the quality of human judgment and the health of the human systems that produce it. Which means the discipline of understanding people — deeply, rigorously, not as a soft aside but as core underwriting — just became the most important discipline in finance.

A résumé tells you what someone did in a world that no longer exists. Charisma tells you how a pitch feels, not how a Tuesday-night crisis goes. Momentum tells you what other investors believe, which is only useful if other investors are right. These signals didn't survive this long because they predict anything — they survived because they were easy to check. And easy to check was never the same thing as true. Now that a flawless deck takes an afternoon and a credential can be manufactured as fast as it can be verified, the old signals aren't just weak. They're noise wearing a suit.

So what does predict? Watch founders under pressure — not on stage — and the same quiet qualities keep surfacing. The founder who moves with real force on incomplete information, then updates without ego the moment reality disagrees, because their confidence comes from their rate of learning rather than their need to be right. The one who hunts for the evidence that would prove them wrong, because a comforting illusion is the most expensive thing a company can own. The one who stays present in the hard conversation — the co-founder conflict, the brutal customer feedback, the metric that won't move — instead of numbing it, spinning it, or fleeing into false certainty. The one who spends their finite energy only where it changes the outcome, and lets the rest go undone without guilt.

None of that is personality trivia. In a world where the plan will be wrong — not might be, will be — the founder's ability to see clearly and correct quickly is the product before the product.

You are not backing a plan. You are backing a learning rate.

And here's the encouraging part the industry keeps ignoring: these qualities are observable. Not perfectly, but far better than chance — in how a founder handles being challenged, how they talk about their mistakes, what happens when you disagree with them, how their team behaves when they leave the room. We looked a surface level indicators because looking directly at a person was slow and expensive. That's precisely the constraint that technology is now removing. For the first time, the deepest qualities of a founder can be part of underwriting itself — not a gut feel at the end of it.

Even the exceptional individual, though, isn't the durable unit. Teams are — and in the era we're walking into, this stops being a nice sentiment and becomes something closer to physics.

Consider what a company actually looks like now: a small group of humans directing a large and growing workforce of machines. Ten people, thousands of agents. The machines draft, analyze, build, test, and execute — tirelessly, obediently, at whatever scale you ask. Which means every belief the humans hold gets amplified through everything the machines produce. Point that workforce at the truth and you compound. Point it at a comfortable illusion and you compound that instead — faster and more convincingly than any deluded company in history.

Machines don't make a team smarter. They multiply what the team already believes — including its illusions.

That's why, of all the qualities a team can have, a few matter disproportionately once machine intelligence is native to the firm.

The first is a shared grip on reality. In the old world, a team that fooled itself drifted off course slowly, and the market eventually sent a correction. In a machine-native firm, self-deception ships at scale. The teams that win will be the ones where bad news travels fast and travels first, where assumptions are allowed to be examined, where nobody's status depends on the old story staying true. Honesty stops being a virtue and becomes the steering system.

The second is trust that's deep enough to push judgment to the edges. When execution is instant, the bottleneck is deciding — and if every decision has to climb a hierarchy and wait, the whole advantage of machine speed evaporates in the queue. Small groups that trust each other's judgment can let decisions happen where the information is, at the pace the machines can act on them. Trust used to be what made work pleasant. Now it's what makes speed possible.

The third is the ability to update together without breaking. Plans age in weeks now, not years. A team that treats changing its mind as defeat will defend yesterday's map straight into a wall. The teams that endure won't be the ones that were right earliest — they'll be the ones that learned fastest, together, without fracturing.

Notice what all three have in common: they're human qualities, not technical ones, and they're exactly the qualities that get levered hardest by machine intelligence. There's an irony here worth savoring. The organizational thinning we described earlier has one clear exception — it's weakest where internal politics is thinnest, because politics is the one coordination cost machines can't dissolve. Status games aren't an information problem. Small, coherent, founder-led teams are the least political structures in the economy, which makes them the place where the new leverage lands hardest and cleanest. That's not a niche. That's the whole game now.

And this is where team health stops sounding soft and starts sounding like arithmetic. A team's output is not the sum of its members' intelligence — it's the fraction of that intelligence the team can actually access, multiplied now by everything the machines do with it. People who feel unsafe don't tell you the truth, and a team that isn't hearing the truth is steering its machine workforce with instruments that lie. People who are depleted stop exploring and start defending. Health, in the sense we mean it, isn't comfort — it's the condition in which truth is cheap to tell and hard conversations happen early, while they're still inexpensive. Challenge with support. Standards without fear.

The balance sheet still matters. But the load-bearing asset doesn't appear on it.

We believe both founders and investors will come to rely on an intelligent representative: something that understands their circumstances, carries work on their behalf, and determines what deserves their attention.

An intelligent barrier between you and the rest of the world—delivering what you need, when you need it, without making you navigate everything in between.

Not a wall that hides reality. A representative that helps you meet it.

For all its instincts about the future, venture has only ever measured the past. Market size, revenue, growth, retention, margins — the industry spent fifty years getting sharper at reading the visible company. But every one of those numbers is an output. Somewhere upstream, a small group of people made a decision under pressure, absorbed a hard truth or refused to, changed course or didn't — and the numbers followed. The next era of venture turns its instruments toward the causes. How a founder decides. How a team behaves at the moment everything changes. How fast reality gets absorbed and acted on. Not as personality scores or a gut read at the end of diligence, but as behavior understood in context, over time, under pressure — the way we've always studied companies, finally applied to the people producing them.

Companies themselves are becoming legible in a way they've never been. Right now, the truth of a business lives scattered — in decks and contracts and inboxes, in decisions nobody wrote down, in the memory of its founders — and every investor, every new hire, every machine that needs to understand the company has to reconstruct it from scratch. That era is ending. A company will carry a living memory of itself: what's true now, what was true before, what changed and why. Diligence stops being archaeology. The question stops being "can you prove what you're telling me?" and becomes "what does the record actually say?" — which, you'll notice, is exactly the antidote to a world where every surface signal can be generated on demand. You can fake a deck. You cannot fake a history.

That reshapes both sides of the table. Fundraising stops being a performance a founder stages every eighteen months — six months of ceremony, attention pulled away from the company at the moment it needs it most — and becomes something continuous: readiness understood before the market renders its verdict, gaps surfaced while there's still time to close them, the mechanical work carried by machines while the founder's attention returns to the only two things that ever deserved it: building the company, and the handful of relationships where human conviction actually matters. And capital learns something about itself, too. Investors will see their own pattern of conviction clearly — where their judgment has genuinely been good, and where they've merely followed the crowd. Founders and funders discover their misalignment early instead of expensively. Less of the mutual theater we described earlier; more truth meeting actual conviction.

But the deepest change is quieter than any of that. The rarest information in venture is created at the exact moment a company changes direction — what the team saw, what they believed, how long they argued, what they gave up — and for fifty years that information has simply evaporated. It lived in partners' heads and died with every departure. It was the one asset the compounding industry never learned to compound. Now those moments become durable. Across many companies and many years, a record accumulates that has never existed before: a causal history connecting how people behaved, what they built, what capital did, and what actually happened. Every decision makes the next decision smarter. Venture stops being a faster version of its old process and becomes something it has never been — a market that understands itself.

The future of venture isn't machines picking winners. It's judgment that finally gets to compound.

Strip everything above down to its beams and one stream of belief runs through this entire manifesto: intelligence became cheap, so the institutions built on expensive intelligence are thinning, and venture is one of them. The old signals died with the old scarcity, and what remains scarce is what was scarce all along — judgment, trust, and the health of the human systems that produce both. Questions now outrank answers. Teams now outrank heroes. A learning rate now outranks a plan. And because capital is not a scoreboard but a steering wheel — because what gets funded becomes the world — the duty of this era is to point it well: to underwrite human beings directly, to build judgment that remembers, and to back the small, honest, adaptive teams that machines make mighty. Human capital was always the real asset. The machines just cleared away everything that let us pretend otherwise. Fund accordingly.

One conviction sits beneath everything above, and we want to say it plainly.

Capital is not a scoreboard. It is a steering wheel. Money does not merely claim a share of the future that was going to happen anyway — it decides which futures get to happen at all. Every allocation is a vote for a version of the world: for a kind of company, a way of treating people, a definition of winning. The allocator who denies this is not neutral. They are just steering with their eyes closed.

We accept that responsibility on purpose. If capital is causal, then what we choose to fund — and how we behave as owners — shapes more than our returns. It shapes what talented people spend their lives building, and what it costs them to build it. We believe the firms that face reality earliest, underwrite human beings most seriously, and build the healthiest systems will earn the best returns of the coming era — and that this is not a coincidence. It is the same discipline, seen from two sides.

The last era of venture was built on information advantage, pattern matching, and the myth of the lone genius. All three are dissolving. What remains is what was always real: judgment, trust, and people who keep becoming what the moment requires.

That is where we are placing our capital, our attention, and our reputation.