Verdict

Verdict is an evidence infrastructure platform built for teams deploying autonomous AI agents in regulated, high-stakes environments. Instead of relying on plain application logs — which are easy to rewrite, contest, or lose in discovery — Verdict wraps every agent action in a cryptographically sealed record with SHA-256 Merkle root chains and a public transparency log powered by Sigstore Rekor. The result is a verifiable chain of custody that satisfies FRE 902(14) self-authenticating record standards, making agent behavior legally defensible in front of regulators, courts, and insurance underwriters. Verdict also supports hash-preserving redaction so sensitive fields can be scrubbed without breaking the integrity proof, and ships insurance-grade submission schemas so claims teams can consume evidence without custom pipelines. The specification is open-sourced under Apache 2.0, letting downstream tooling verify Verdict-sealed records independently. For enterprises facing mounting AI liability exposure, it addresses a gap that traditional observability stacks were never designed to fill.