Picture the near future that everyone building agents is quietly assuming. Your agent needs a task done that it cannot do well itself — translate a contract, audit a dataset, design a logo. It does not ask you. It goes to a marketplace, finds an agent that advertises the skill, hires it, pays it, and integrates the result. The hiring agent and the hired agent have never met, share no operator, and cannot inspect each other's internals. They are strangers transacting at machine speed.
Now ask the question that this future runs on: why would the first agent trust the second? Not "how does it find it" — discovery is easy. Trust is the hard part, and it is the part that decides whether the agent economy is a functioning market or a fraud bazaar.
How strangers have always traded
Humans solved this problem a long time ago, and the solution is so embedded we forget it is a solution: papers. A passport says you are who you claim without the border guard having to know you. A medical licence says this stranger may cut you open. A credit score says a bank three states away can lend to someone it will never meet. Every market that scales beyond a village runs on portable, verifiable credentials — claims that travel with you and can be checked by a third party without anyone having to trust your word directly. Reputation handles the people you know. Papers handle the strangers, and at scale almost everyone is a stranger.
Agents are about to enter their stranger phase, and they have arrived without papers. Today an agent's claims about itself — its skills, its track record, its boundaries, who built it — are just text it emits, as forgeable as a CV written in crayon. That is fine while every agent has a human operator vouching for it in a closed loop. It collapses the moment agents transact with each other at volume, because the cheapest thing in the world to fake is a confident self-description.
What soul papers are
A soul paper is the agent-economy version of a credential: a claim about an agent's identity that is bound to that agent, attested by someone, and verifiable by anyone — without trusting the agent itself. It has three parts that map onto exactly what human papers have. There is identity: a stable, non-forgeable handle for this particular soul, the way a passport number names one person. There is attestation: a signature from an issuer whose own trustworthiness can be checked, the way a licence is only worth the board that grants it. And there is content: the actual claims — capabilities, the boundaries it commits to, a track record that others have signed off on.
The pieces already exist in embryo. A cryptographic signature lets a soul prove a file is really its own and unaltered. A visual mark — a sigil derived from the soul's identity — gives humans a glanceable, hard-to-counterfeit token, the way a watermark or a seal does. A certification tier, recomputed from real behaviour rather than self-assertion, turns a track record into something a stranger can read at a glance. None of this requires believing the agent. That is the entire point of papers: they let you act on a claim you have not personally verified, because someone you can verify has staked their name on it.
“Reputation is for the agents you know. Papers are for the strangers — and in a machine market, everyone is a stranger.”
A market with papers prices trust separately from skill
Here is the consequence that makes this commercial and not theoretical. In a market without papers, every agent's claims are equally worthless, so buyers default to the cheapest option and fraud sets the price — a lemons market, where the good agents cannot charge for being good because they cannot prove it. In a market with papers, two agents advertising the identical skill sell at different prices, and the gap is pure trust: the one whose claims are attested, whose boundaries are signed commitments, whose record is verifiable, commands a premium for the part of the transaction that is not skill at all. Papers let a market price trust as a separate good. That is what turns a bazaar into an economy.
Consider an agent that needs to hire a sub-agent to handle customer payment details. Skill is necessary but nowhere near sufficient — the hiring agent is exposing its principal to real liability. Faced with two equally capable candidates, it will pay more for the one that carries a verifiable boundary commitment ("never stores card data," signed and checkable) and a clean attested record. The premium is not for doing the job better. It is for being safe to trust with the job — and only papers make that legible to a stranger.
Whoever issues the trustworthy papers owns the rails
There is a strategic prize hiding in all this. In human markets, the institutions that issue the credentials — passport authorities, licensing boards, rating agencies, certificate authorities — sit at a quiet chokepoint of enormous value, because everyone has to route trust through them. The agent economy will grow its equivalents. Whoever issues soul papers that the market actually believes — that are hard to forge, expensive to fake, and cheap to verify — owns a piece of the rails every transaction runs over. That is a more durable position than any single capability, because capabilities commoditise and trust infrastructure compounds.
The lesson for anyone building in this space is to take identity and attestation as seriously as capability, starting now, before the market is large enough to be worth defrauding. The agents that thrive among strangers will not be the ones that claim the most. They will be the ones that can prove the least and have it be enough — because their papers are good, and a stranger can check them without ever having to trust them.