Founder of Neura — the authority and evidence layer for consequential AI actions.

Neura sits between an agent’s proposed action and execution:

proposed action → authority decision → execute or restrain → Decision Receipt

I work with organizations and builders that need high-impact agent actions to be explicitly authorized, bound to the exact target, and independently verifiable.

• Evaluate Neura — bring one consequential company-owned action into a bounded evaluation.
• Build with Neura — run a local check and connect an agent runtime.
• Relay Action Card — explore the public schemas, examples, adapters, and verification path.

• NVIDIA: NeMo Agent Toolkit and OpenShell review threads on invocation-bound HITL context, caller contracts, JSON-RPC/MCP policy, and auditable execution boundaries.
• Microsoft / Azure: Azure Cosmos DB Shell and Semantic Kernel work on destructive-action confirmation, fail-closed authorization, and exact-call execution evidence.
• Anthropic ecosystem: Claude Cookbooks #701 explicitly credited my executor-side exact-action framing; community PR #803 implements and tests digest binding, single-use approval, replay refusal, and execution receipts in an Anthropic-owned repository.
• OpenAI / Codex: Codex #31565 boundary and test-shape feedback on delegated MCP approvals, answerable authority, and terminal no-effect outcomes.

See the source-backed evidence index →

These are public technical contributions and ecosystem signals—not claims of endorsement, partnership, adoption, integration, customer status, or official validation by the named organizations.