Fail-closed accountability layer for agentic AI: eight versioned, evidence-fed witness assessors evaluate an agent's proposed action before it executes; every judgment is sealed and written to a tamper-evident hash-chained attestation record, with assessor dissent preserved rather than averaged away.

Fund this project

Unverified URL

The funding manifest has not provided proof via wellKnown that this link is associated with it. Learn more.

Continue