The Manav Manifesto

We didn't build a login. We built the cryptographic claim that you, the human, were here — and that what was done in your name can be proved to the world, traced through every machine that helped, and paid back to you in proportion to what you actually contributed.

One — The trust layer is the era's defining problem.

Every era of computing has had a moment when the question of trust ceased to be optional. DNS in 1983, when "which server" outgrew the hosts file. SSL in 1995, when "which connection" outgrew the open wire. OAuth in 2007, when "which app" outgrew shared passwords. Each was an answer to a question the prior decade had failed to ask cleanly. The answers became plumbing. The plumbing became civilisation. Anyone arguing that the trust layer is "just infrastructure" forgets that infrastructure is the only software whose builders end up named in textbooks.

Two — The new question is older and newer than any of these.

The question is which human, and it is older than any computing question because we have asked it for ten thousand years across every culture: who can be trusted, who is accountable, who is real. It is newer because, for the first time, it is no longer rhetorical. It is an engineering problem, a legal problem, a regulatory problem, a $2.1-trillion-a-year economic problem. The answer it deserves is engineered, not asserted.

Three — The actors on our networks are not, mostly, human.

For the first time in history, the entities most active on the network are not human. The average enterprise manages 250,000 non-human identities and a few thousand human ones. The ratio inverts the assumption every prior identity protocol was built on. Login flows assume one human. Audit logs assume one user. Anomaly detection trains on user-shaped behaviour. None of those assumptions hold any more, and patching them in place has run out of tape.

Four — A login confirms a session. We needed something larger.

So we did not build a better login. A login confirms that someone with a password — or a passkey, or a passing iris scan — is at a keyboard at a moment. We built a protocol that anchors every action to a verified human root: every agent decision, every approved transaction, every shipped piece of work, every credential issued and every credential consumed. Not at one moment, but as a continuous chain whose links can be reconstructed by any party with reason to ask.

Five — We chose a name older than every machine ever built.

We chose a Sanskrit word — मानव, "human" — because we wanted a name that did not require translation. 1.4 billion people already speak some descendant of Sanskrit; the other 6.6 billion can pronounce it on first try. The agentic machine deserves a counterweight whose name predates the wheel, the alphabet, and the city. We are not the inventors of the human; we are stewards of a protocol named after one.

Six — The protocol has five layers, each answering a question the prior decade ducked.

Identity. Are you human, and are you the same human you were last week? Delegation. Did you authorise this agent, and within what scope? Attestation. Did you do this work, supervise it, or merely direct it? Trust. Has the network seen you do good work before? Settlement. Will the network pay you for what you contribute? Skip any one and the stack collapses into a partial answer that fails its first hard audit.

Seven — We open-sourced the core because trust layers cannot be owned.

The protocol is Apache 2.0. It is not ours to keep. The trust layer of the agentic age, like the trust layers before it, must be infrastructure nobody owns and everyone can run. We will earn our place by shipping the best implementation, not by erecting the highest fence. The day someone forks Manav and runs it better, we will have done our job.

Eight — The token is the gas, not the prize.

$MANAV is not the point of this project. It is the economic mechanism that makes the protocol self-sustaining without surveillance. It is the gas that powers verification, the receipt that says a real human did real work, and the rail that pays them for it. It is mined by working, not by burning electricity, not by buying low and waiting. The token's value tracks the network's verified human contribution. We engineered it that way on purpose, and we will defend that property through every iteration of the protocol.

Nine — We will not sell user identity data. The protocol enforces this.

Not as a marketing claim — as an architectural property. Selective disclosure is the default, not the upsell. The user reveals the minimum claim required, in zero-knowledge form, with the underlying data never leaving their device. If we ever changed that posture, the protocol would stop working in our favour before it stopped working for users. We are constrained by the same cryptography that constrains every other party in the system, and we like it that way.

Ten — We are not against AI agents. We built this for them.

An agent without a human root is a liability for everyone in its blast radius — the operator who cannot audit it, the regulator who cannot sanction it, the insurer who cannot price it, the customer who cannot trust it, the human whose name it acted in. Manav is what makes agents more useful, not less, because it makes them legible to the systems that need to grant them authority. We are the protocol the agent economy will look back on as the one that let it grow up.

Eleven — We will be wrong about many things.

The emission curve will need adjusting. The trust score will be gamed in ways we did not anticipate. The first attestation schemas will be replaced by better ones. The threat model will discover gaps we did not see. We will publish corrections, not press releases. The credibility of the protocol depends on the honesty of the project that builds it, and on every step of the way looking less like marketing and more like science.

Twelve — Here is what will not change.

The human is the root of all trust. Every machine on the network derives its authority from a human who delegated. Every output the network values can be traced back to a human who contributed. Every transaction that needs to mean something needs a human who can be held accountable. Strip these away and you have not the agentic age but its caricature: a hall of mirrors where everything signs everything and nothing means anything.

We refuse the hall of mirrors. We choose the cryptographic claim. The human was here. The human authorised this. The human did this work. The human is real, and accountable, and worth paying.

That is the protocol. That is the company. That is the manifesto.

The human is the root. The agents are the branches. Manav is the trunk.

Signed and on-chain by the founders. The hash of this document is anchored at manav.id/manifesto.sig. If the words above ever change in a way that contradicts the original signature, that change will itself be cryptographically visible. We are bound by the same machinery the rest of the protocol is bound by.