Action State launches AI verification tools

Action State Group Inc. launched open-source tools for businesses to independently check ai agent interactions without central authority.

Tools include EvidenceBook, Terms Packs, Countersign

EvidenceBook preserves tamper-evident records, Terms Packs evaluate agent behavior, and Countersign provides independent verification services.

Founder explains key difference

Steven Mih, Action State's founder, said the tools differ from observability vendors by focusing on cross-company transactions rather than internal systems.

“is that most of the observability and tracing has to do with what was happening within your own walls”

EvidenceBook works offline and is free

EvidenceBook allows participants to share signed transaction histories without central authority, and verification occurs offline at no cost.

Countersign checks cryptographic properties

Countersign verifies cryptographic and structural properties of evidence bundles but does not certify outcomes or completeness.

Terms Packs address regulatory compliance

Terms Packs define evaluation criteria for agent work against regulatory rules, with an initial pack covering eu ai act obligations.

Human review critical for accuracy

AI judges assess agent records daily, with employees periodically reviewing samples to compare human and automated judgments.

Early integrations include Mesh-LLM and Project Nanda

Mesh-LLM and Project Nanda use Agent Action Capsules for sealed bilateral records in ai agent transactions.