Building the future of autonomous AI systems
As Chief Platform Officer at Opaque Systems, I build open standards for running autonomous agents with verifiable, hardware-attested trust. My current work lives in the agentrust-io project, and I created the Agent Governance Toolkit hosted under Microsoft.
Trust Runtime Attestation and Compliance Evidence. A standard for hardware-attested trust records that a verifier can check without trusting the operator running the agent.
Confidential MCP gateway. Every Model Context Protocol tool call is evaluated against a Cedar policy bundle inside a hardware TEE, with the bundle hash measured into the attestation report and a signed proof produced per session.
A cryptographic identity and provenance standard for AI agents. It hardware-anchors the artifacts that define an agent at deployment, so what is actually running can be proven rather than asserted.
Confidential agent-to-agent. Attested, attenuated delegation and sealed peer channels expressed as a profile on the A2A protocol.
The open-source platform I created for runtime governance of AI agents, now hosted under Microsoft. Policy enforcement, zero-trust identity, execution sandboxing, and reliability engineering for autonomous agents.
These began as standalone projects and were folded into the Agent Governance Toolkit. Their original repositories are now marked deprecated and point to AGT.
The policy engine and governance kernel: deterministic enforcement with POSIX-inspired primitives and a layered privilege architecture. Now the policy core of AGT.
Zero-trust identity mesh for AI agents: Ed25519 identity, trust cards, and cross-org federation. Now the identity layer of AGT.
Reliability engineering for AI agents: SLO engine, chaos testing, deterministic replay, cost guard, and progressive delivery. Now the reliability layer of AGT.
A written guide to production agentic system design: the Inference Trap, the Guardrail Router, and the compute-to-lookup ratio.
My methodology for reliable AI systems: remove complexity rather than add features. Control planes over prompts, graphs over context, silent swarms over chat.
A curated list of tools, frameworks, standards, and resources for AI agent governance, safety, and compliance.
The Agent Governance Toolkit ships governed examples and adapters for around ten agent frameworks, including LangChain, CrewAI, AutoGen, Google ADK, OpenAI Agents, LlamaIndex, and smolagents, alongside MCP and A2A. See the toolkit repository for the current, authoritative list.
Framework integrations on GitHub →Check out my open source contributions and experiments on GitHub
View All Repositories on GitHub →