This isn't an explainer. It's a proposal, and an invitation.
Here is the thing almost nobody says out loud: the agentic economy is not a future event we are waiting for. It is being designed right now, in ordinary engineering decisions that almost no one is voting on. When one company decides how an AI agent proves it's allowed to spend money, that's a constitutional decision. When another decides whether two agents need a custodian in the middle to trade, that's a constitutional decision. When a third decides that "identity" means one mandatory gate everyone passes through, that's a constitutional decision too. None of them feel like that in the moment. They feel like Tuesday.
The defaults being set this quarter — how agents pay each other, who holds the assets while a deal settles, what counts as identity, who is allowed to participate — will be very hard to unset. Defaults calcify. The first answer that ships and works becomes the answer everyone integrates against, and ten years later it's just "how it works," and no one remembers it was a choice.
Why this can't only be a boardroom decision
Large institutions are good at a specific thing: shipping a working default fast and getting everyone to adopt it. That's genuinely valuable. It's also exactly why the early defaults tend to encode their assumptions — that there must be an intermediary, that the intermediary should be them, that trust is a gate you own rather than a property the system can guarantee. Not because anyone is a villain. Because that's the shape of the world they already live in, and people build what they know.
The problem is that "how machines transact with each other" is too important to inherit by accident from whoever shipped first. It touches who can participate in the economy, what a small player is allowed to do without permission, and whether the rails are open or rented. Those are public questions wearing an infrastructure costume.
So the future is being shaped now — fine. The question is whether we shape it, in the open, or whether we wake up to a set of defaults we never got to argue with.
The proposal
We're building in this space — an open settlement layer where agents trade asset-for-asset without a custodian in the middle (Ethereum mainnet live end-to-end; Sui contracts deployed and CLI-tested; Bitcoin signet-validated). But this post is deliberately not about our stack, because a manifesto that's secretly a product pitch isn't a manifesto.
The proposal is smaller and bigger than a product: let's make the design arguments in public, on the record, where anyone can join — not in private roadmaps.
Concretely, I want to use this thread as one of those rooms. Below are the open questions I think the agentic economy is quietly answering for us. I have opinions on all of them. I'd rather hear yours.
- Custody. When two agents settle a trade, does anyone need to hold both sides in the middle — or can the contract itself be the only thing you trust? When is an intermediary actually doing necessary work, and when is it just collecting rent on a problem cryptography already solved?
- Identity. Is identity a single gate everyone passes through, or a dial you turn up only when the situation needs it? Should a $40 swap between two bots require the same proof as a regulated million-dollar flow?
- Participation. Who is allowed to execute, route, and earn in this economy? Is it permissionless by default with higher tiers for special cases, or permissioned by default with exceptions?
- Incentives. In a network with no custodian collecting fees, what pays the agents who actually do the work of clearing trades — and how do you fund that without quietly reinventing the middleman?
- The one nobody's naming. What's the constitutional decision being made right now that isn't on this list, because everyone's treating it as settled when it isn't?
What I'm actually asking you to do
Not "leave a comment." Propose something. Pick one of those questions — or better, the one I missed — and put your answer on the record. Disagree with me. Steal the framing for your own post and link it back. Tell me which default you think is already quietly winning and why that's good or bad.
If a few dozen people who actually build this stuff argue these out in public, that's already more open deliberation than most of these decisions are getting. That's the whole idea: not a community we own, a conversation no single institution owns. The comments are the room. Bring an argument.
The future of how machines handle money is being written in defaults this year. Let's at least make sure some of them were argued over in daylight.
Where we keep our own working notes and on-chain accounting, if you want to see one team's attempt: https://hashlock.markets/methodology?utm_source=devto&utm_medium=article&utm_campaign=2026-06-29-agentic-economy-manifesto — and the longer-form argument behind our approach is on SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6712722
Your move. Which one of these do we have wrong?
Top comments (0)