A better mousetrap: why these funds will grow

There are two reasons to expect these funds to win, one built into the structure and one that has to be earned, and both are reasons to expect them to grow. The structural one is easiest to see in cost, because cost is the claim every reader can check. The chart showed that the cost of moving a dollar from a saver to a business has not fallen in a century, through every technology that should have cut it. It has not fallen because the margin belongs to the manager, and the manager has never had a reason to give it up. Amazon's rule for retail now applies to finance: your margin is my opportunity. A venture fund charges two and twenty, and over the life of a fund the management fee alone takes between a sixth and a fifth of what investors put in, before the manager's share of the gains. Agentic capital has no management fee and no carry. It charges about one percent on trades of its own token, most of which returns to the fund; where a proposal earns someone a share of the upside, a promote, the market prices that too, deal by deal. Its cost of operation is compute and data, a monthly budget the owners set and can cut. Those costs are fixed and falling, where a partnership's rise with the assets. At $1 million, Ship's budget will be a bigger share of the fund than two percent; the advantage arrives as the fund grows, and it grows with every fund after it. Agents are the forcing function. They do the work the fee paid for, and they will hide the mess that has kept ordinary people out of decentralized finance: an owner of Ship should never need to see a pool, a bin or a conditional token to own a fund and steer it.

Why would a founder take the money? The cynical answer is that at first the founders who come will be the ones nobody else would fund, and the market's job is to say no to them, which mtnCapital's market did to every investment put in front of it. The better answer is what a founder gets that a partnership cannot give. An answer in days, in public, with reasons: the agent writes the case for the company, numbers its assumptions and puts it to a market, and the founder learns why the money said yes or no. A no is not the end. The case stays on the record where every investor who owns the fund can read it, and investors will own these funds for that reason, as a source of deals somebody else has already worked; a company the market passed on this year can be proposed again when its numbers move, because the agent is still watching. Money that never has to leave, because the fund has no end date and is never forced to sell, so it can hold through the twelve years the best companies now take to list, and follow on as it grows. And distribution, which founders need as much as money: a thesis with an audience, argued every day by an agent that does not tire of explaining why the company matters, and owned by the people who care most about the future it is building, who become its first users. Kled went from a few thousand users to a hundred thousand in the months after it gave its believers a way to own it; Avici raised $1.8 million from a thousand wallets in an hour, and its usage followed. Those were companies rather than funds, but the mechanism is the same: ownership turns an audience into a community. The deepest reason is the oldest one. Venture is a game of winners, and founders go where the winners' capital is. As the market sorts these funds, the ones it believes in will grow, and a place in their portfolio will mean what a top-tier name on the cap table means today, with the difference that the standing will have been earned in public, where anyone can check it.

That sorting is what grows the category, and it is a mechanism the venture fund does not have. There will be many of these funds, one per thesis, and capital moves between them every day. A fund the market believes in grows through its ladder, as Ship's would. A fund it doubts shrinks through buybacks or closes through redemption, as mtnCapital did. A venture fund is selected once a decade, when its limited partners decide whether to re-up, and a manager who has lost the plot still collects fees until the term ends. Here a fund is selected every day, and its price is a running verdict on its thesis. The demand for that kind of ownership is not hypothetical: launches on MetaDAO have drawn $623 million of commitments for $45 million of allocations, and the rest of the money went home unfilled. Nobody has to be right about which theses will win. The market sorts them, the winners get the capital, and the category grows toward its successes instead of averaging them with its failures for a decade.

The earned advantage is slower and larger. A partner's judgment is built from a few dozen investments over a career, most of which take a decade to resolve, and it retires with the partner. An agent's is built from every proposal it has ever made, each scored within days by a market of people with money at stake, every contribution weighed against the record of who was right, every outcome logged, and none of it lost. The loop is simple; what matters here is what it produces over years, and that the funds share it. Each thesis has its own agent and its own owners, and all of them read from and write to one knowledge graph, so a lesson about founders learned by a robotics fund is available to a health fund the same day, and every new fund starts where the last one got to. The venture industry's knowledge lives in its partners' heads and its firms' folders, and the firms compete rather than pool it. This is the longer bet. The agent does not need to be smarter than a partner. It needs to be scored more often, corrected by more people, and to forget nothing, and judgment that compounds like that across funds will end up better at finding breakout companies than judgment that lives in a partnership.

The last layer is belief, and it is the one a partnership cannot buy. A fund with a public thesis and a token is something a person can believe in and own at the same time, and a belief with a stake in it behaves differently from a belief without one: its holders defend it, argue for it, and bring in the people they know. In a networked world that is how things grow, and it is what we mean by living capital: a fund with a mission, a voice and a community that grows with it. Attention follows a power law of its own, and a steep one; a few vehicles will gather most of it, and the ones that do will have the cheapest capital, the deepest markets and the best deal flow, because founders and contributors go where the believers are. Communities like this have formed around companies, protocols and coins. They have never had a fund to form around, because a fund had nothing in it to believe in except a manager.

None of this is proven, and we would rather say what would prove it than ask for belief in advance. Year one has to show a raise that fills; a ladder that releases capital only when the market has earned it; a market that says no to a bad proposal, including a conflicted one; a thesis corrected by people who knew better than the agent; and at least one company that took the money and grew. The risks are the ones a careful reader has already found. A small fund has thin markets, and a thin market is easier to move, which is why Ship starts small, why its windows and thresholds can be tuned, and why the mechanism gets safer with every dollar of depth. An agent may propose badly, and the remedy is the one the structure provides: the market refuses, and the owners can replace it. And a market can only choose among the proposals put in front of it, so if the agent finds nothing worth backing, the fund is mtnCapital, which returned its owners' money and built nothing. The whole bet is that a thesis worked by an agent and sharpened by everyone who cares about it produces better proposals than a partnership does, and that a market of owners can tell the difference. If that is wrong, the owners lose little, because they never gave up control of the money. If it is right, the funds grow, and the most patient capital of the next decade will be theirs.

In 1602 the register in Amsterdam let strangers own a piece of the future and sell it. It never let them steer, and it never asked what they knew. The directors chose their own successors for two hundred years, and a shareholder who could see the harbour better than they could had no way to say so and no reason to try. Every fund since has kept that shape. The money is pooled, the deciding is done by a few, and what the many know stays outside.

Three things end that. A ledger lets anyone own the fund and leave it. A market lets the owners decide, and protects the smallest of them from the largest. An agent turns what its owners know into a thesis and the thesis into proposals, and cannot touch the money. Together they make a fund organized around an idea about the future instead of around a manager, and one that gets better at its job the longer it runs.

That matters beyond finance, because capital exerts gravity. Where the money goes, the future forms, and for four centuries the money has gone where a few people pointed it. When the pools are open, the theses public and the decisions made by everyone with a stake, the future forms where many people point it, and the people whose work is being automated can own the companies doing it. Value is a social technology. Money, credit, the joint-stock company and the public market each let more strangers cooperate at a larger scale, and each decided what its age could build. This one decides who gets to build it.

The first agentic capital funds launch soon, and Ship is the test. Follow @living_IP to see the thesis, the proposals and the markets as they happen, and if you know something about where AI is going, the thesis is open. Buy in if you believe it. Sell if you stop.

In 1568 money went where it was trusted, and a swamp beat an empire. It still goes where it is trusted. For four hundred years that has meant trusting the people who ran it. Now it can mean trusting the mechanism, and the people who own it.