Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces Governance-Aware Agent Telemetry (GAAT), a reference architecture for real-time policy enforcement in multi-agent AI systems. GAAT extends OpenTelemetry with a Governance Telemetry Schema (GTS), a low-latency policy violation detection engine, a Governance Enforcement Bus (GEB) for graduated interventions, and a Trusted Telemetry Plane for cryptographic provenance. By closing the loop between telemetry collection and automated policy enforcement, GAAT enables proactive governance and prevents policy violations before damage occurs.
Stop playing catch-up: GAAT lets you enforce multi-agent AI policies in real-time, not just observe violations after they happen.
Enterprise multi-agent AI systems produce thousands of inter-agent interactions per hour, yet existing observability tools capture these dependencies without enforcing anything. OpenTelemetry and Langfuse collect telemetry but treat governance as a downstream analytics concern, not a real-time enforcement target. The result is an"observe-but-do-not-act"gap where policy violations are detected only after damage is done. We present Governance-Aware Agent Telemetry (GAAT), a reference architecture that closes the loop between telemetry collection and automated policy enforcement for multi-agent systems. GAAT introduces (1) a Governance Telemetry Schema (GTS) extending OpenTelemetry with governance attributes; (2) a real-time policy violation detection engine using OPA-compatible declarative rules under sub-200 ms latency; (3) a Governance Enforcement Bus (GEB) with graduated interventions; and (4) a Trusted Telemetry Plane with cryptographic provenance.