Flight Recorder

Agent Flight Recorder

Agent Flight Recorder is the black box for AI agents.

We were inspired by a simple shift in enterprise AI: agents are moving from answering questions to taking actions. Once an agent can send an email, update a CRM, change a work order, deploy code, prepare a payment, or expose sensitive data, companies need more than chat history. They need governance before execution and proof afterward.

Agent Flight Recorder sits between an AI agent and the company tools it wants to use. Every proposed tool call is captured, evaluated against policy, and either allowed, blocked, or routed to a human approver.

The core flow is:

Agent request -> Flight Recorder gateway -> Policy check -> Human approval -> Tool execution -> Evidence packet

We built the project as a live enterprise action-governance demo. It includes:
A React/TypeScript control console
Netlify Functions for gateway and approval logic
Policy-based risk classification
Human approval gates
Timed step-up approval for high-risk actions
Slack-style approval routing
ElevenLabs voice review
Exportable evidence packets with audit hashes
A simple way to describe the product is:

Trust = Visibility + Control + Evidence

The biggest thing we learned is that "human in the loop" is too broad. The valuable control point is the action boundary: the exact moment before an agent touches a real system.

The hardest challenge was making governance feel operational instead of abstract. We wanted the demo to show what is happening in real time: the agent action path, the policy gate, the approver decision, and the evidence packet that proves what happened.

Agent Flight Recorder is built for a future where agents do real work across finance, civic services, healthcare admin, supply chain, media rights, customer operations, and software teams. The mission is simple: let agents move faster while giving companies the control and proof they need to trust them.

Built With

• agents
• audit
• automation
• compliance
• cybersecurity
• edge
• elevenlabs
• engine
• functions
• human-in-the-loop
• javascript
• management
• policy
• postgresql
• react
• risk
• slack
• supabase
• typescript
• vite