Loading…
Trust Center
Which outputs a model may produce alone, and which require a qualified reviewer — stated plainly, because that boundary is what makes an output trustworthy.
Verassa evidence protocol
Evidence
01Screenshot, DOM, replay, and axe baseline captured before decisions.
Judgment
02Reviewer route, rationale, and owner stay attached to lower-confidence work.
Verification
03Re-scan records and disclaimers travel with reportable outputs.
The gate
Agents do the production work: evaluating a scope, capturing evidence, drafting findings. That work can become an internal working document without a reviewer in the loop, as long as it is clearly labelled as a draft.
Any output a third party will rely on is different. An external report, a VPAT or ACR draft, a defense documentation package — each requires a named qualified reviewer before it leaves the building.
Report states
A draft internal report carries no external reliance and is generated with a clear draft label. A review-ready report has findings and evidence assembled and waiting for a qualified reviewer. An external-reliance report has been reviewed, carries a named reviewer, and includes the required disclaimer.
A report does not silently move from draft to external-reliance status. Promotion is an explicit step, with a person in it.
Accountability
An external-reliance output records who reviewed it and when. A reviewer's name on a deliverable is not decoration — it is the accountability that makes the deliverable defensible.
This is also why the platform will not generate a conformance attestation from agent output alone. Conformance is something a qualified person attests to; the product assembles the evidence that makes that attestation possible.