The proof layer for agentic commerce
AI agents are transacting on the web. x401 lets any service ask an agent “can you prove who's behind this request?” — over standard HTTP — and get a verifiable credential back before it grants access.
Every major agentic commerce framework — Visa TAP, Stripe ACP, Mastercard Agent Pay, Google AP2 — defines what a credential can do. None define who is behind it. When an agent executes a financial transaction, there is currently no mechanism to prove a verified human authorized that specific action, under specific constraints, at that specific time.
x401 defines four nodes in the authorization chain. Each node has a verifiable artifact. Remove any node and the chain breaks — the transaction cannot proceed.
x401 rides on plain HTTP — no new transport layer to adopt. Three exchanges between the service and the agent's wallet, carried in standard headers, complete the entire proof handshake.
Before it grants access, a verifier returns a small JSON payload in the PROOF-REQUIRED header — naming the credential the agent must present to prove the verified human behind it. Nothing hidden in a proprietary format. Here's the canonical example from the spec.
x401 is being developed with organizations actively deploying agents at scale across payments, insurance, and financial services. Formal announcements are coming. The spec is open now — these use cases illustrate why early implementers are moving fast.
x401 is an open protocol — any CA can issue conformant credentials. Proof is the primary implementer: a WebTrust-certified CA with FIPS 140-2 Level 3 infrastructure, Kantara IAL2 enrollment, and an 8,000+ relying-party network. Build on Proof and your agents start with credentials that are trusted by default — no onboarding friction for your users.
The spec is open. The reference implementation is live. Leading organizations in payments and enterprise insurance are already building on it. The window to establish the proof layer for agentic commerce is now.