Web proof verification

Most Vouch verifications use this mode. verificationMethod defaults to "web-proof", so there is nothing to set.

What the user does

What you receive

The same payload shape as every other Vouch verification: a requestId, the outputs the data source defines, and a webProofs array carrying the evidence.

{
  "requestId": "…",
  "verificationMethod": "web-proof",
  "outputs": { /* the facts you asked for */ },
  "webProofs": [
    {
      "outputs": { /* extracted from this proof */ },
      "presentationJson": { /* the evidence, re-verifiable by anyone */ },
      "decodedTranscript": { /* human-readable request and response */ }
    }
  ]
}

You do not have to do anything with webProofs. Vouch checks every proof before the webhook is sent, so the outputs are already trustworthy. The evidence is there for the cases where you need to audit a result later, or let a third party confirm it independently. Receiving and verifying results covers both.

What makes it trustworthy

Vouch uses zkTLS and confidential computing to attest the platform's response. The user's session is not replayed, simulated or scraped: the proof is made from the live connection between the user and the platform, so a doctored screenshot or an edited response cannot pass.

That machinery runs underneath the product. You consume a normal data API and never operate any of it, and your users need nothing beyond their own account.

When a source uses video instead

Some things a user can clearly demonstrate on screen cannot be captured as a web proof. Those sources use video verification instead, where the Vouch app screen-records the session and an AI verifier returns the same structured outputs.

This is decided per data source when the source is built, not per request. The catalog marks which mode a source uses, and passing verificationMethod: "video" to a source that does not support it is rejected.