Three filings.
One standard.

Abstract

A system and method for capturing cryptographically sealed evidence records of AI agent actions that render into multiple regulatory formats — SOC 2 dossiers, EU AI Act Article 12 logs, FRE 902(14) self-authenticating court exhibits, and insurer underwriting submissions — without rework. One sealed record satisfies simultaneous compliance with auditing, regulatory, evidentiary, and insurance standards.

In plain English

Verdict captures what autonomous AI agents do at runtime and seals it cryptographically so the evidence cannot be modified. The sealed record is smart enough to render into whatever format regulators, courts, insurers, and auditors need — without rework.

Core invention pillars

• ◈Closed taxonomy of canonical AI agent event types
• ◈Content-addressed policy bundle versioning via SHA-256
• ◈RFC 6962 Merkle trees + Sigstore Rekor anchoring + Ed25519 HSM signing
• ◈Hash-preserving redaction (GDPR Article 17 compliance without breaking proof)
• ◈Evidentiary Completeness Score (ECS) → FRE 902(14) certification map

Abstract

A system and method for pre-execution verification of autonomous agent intent against a hierarchical policy tree, enforcing graduated gatekeeping decisions (allow, require_human_review, escalate, or deny) at the atomic action level, providing cryptographic evidence sealing of all policy evaluations, and enabling deterministic runtime governance with formal verification of policy conformance.

In plain English

Before an AI agent acts, Guardian checks whether the action is allowed under the current policy rules. Allowed → execute. Questionable → human review. Forbidden → block. Every decision is cryptographically sealed and timestamped with proof of which exact policy rules were in force.

Core invention pillars

• ◈Atomic decomposition of multi-step instructions into primitive actions
• ◈Hierarchical policy tree with O(1) compliance checking
• ◈Three-state gating with cryptographic evidence of evaluation
• ◈Real-time policy bundle version reference in every event

Abstract

A system and method for detecting behavioral drift in autonomous agents at runtime using embedding-distance analysis, triggering graduated intervention (alerting, fallback mode, token revocation, kill switch) with sub-50ms latency, cryptographically sealing all intervention events, and enabling dynamic capability-based access control revocation to gracefully degrade distributed agent pipelines during anomalous behavior.

In plain English

During execution, Circuit continuously watches the agent's internal decision vectors. If the agent starts drifting from trained behavior, Circuit intervenes automatically — first warning, then safe-mode, then revoking permissions, then killing the agent if necessary. All in under 50ms. Every intervention is sealed as cryptographic evidence.

Core invention pillars

• ◈Embedding-drift-triggered classification across four graduated tiers
• ◈Sub-50ms intervention via lock-free ring buffer + signed capability tokens
• ◈Dynamic CapBAC permission revocation
• ◈Cryptographic sealing of every intervention in the Merkle tree