The agent era needs a new identity layer.

Every inbox you own is filling with messages from things that aren't quite people. LLMs sign emails now. Customer-service "humans" are scripts. Cold outreach is AI-drafted at industrial scale. The sender field is a lie, and it has been for years — the LLM era just made the lie cheap enough to be everyone's problem.

One handle. Verifiable. Yours.

A USERNAME.md handle is /u/yourname/ — a cryptographic identity rooted on a real domain. The math behind it is open standards: did:web, JSON Web Signatures, and W3C Verifiable Credentials. The brand is ours. The choice of who issues claims about you is yours.

When an agent reaches out on your behalf — to book a flight, schedule a call, complete a purchase — it can prove who sent it. When an agent reaches out to you, your inbox can prove who sent that. Same crypto, opposite direction. Forged messages don't get a green check; they get auto-quarantined before you read them.

A three-sided system.

USERNAME.md is the subject side: where individuals get claims attached to them. ABOUT-ME.md is the long-form profile that anchors those claims to a person. ABOUT-US.md (coming next) is the issuer side: where organizations issue signed credentials about people they know. And the interaction surfaces — outreach.md for cold introductions, sink.md for context, minder.md for the things you don't want to forget — are where claims show up and decisions get made.

This is the substrate for agent commerce. Not because we own the agents — because every agent that wants to transact needs to prove who it represents, what it's authorized to do, and what it's promising. That requires identity, delegation, and receipts. We ship all three.

Why now, why us.

Because email and DMs were never identity layers, and now everyone is finally paying for that. Because Bluesky, AT-proto, ENS, and Keybase each ship a piece of this but nobody ships the consumer outcome: an inbox where the green checkmark means something. Because the alternative is letting Google and Apple decide which AI agents are allowed to reach you — and they won't, because the honest answer cannibalizes their ad model.

We're betting the agent era will need its own user-controlled identity layer. We're building it — on standards, on domains we own, with refunds if we don't ship.