AI Sandbox
A public, no-KYB environment that hands you a governed agent — first allow / observe / block / escalate in minutes.
What it does
The AI Sandbox is a public, no-KYB environment that hands you a pre-issued governed agent so you can see a real allow / observe / block / escalate decision in minutes. It runs the same policy engine and the same code path as production — the only difference is that you graduate to your own verified identity and mandates when you go live.
Inside AI Sandbox
The core functions this engine provides across the Trust Fabric.
AI Sandbox composes with the Trust Fabric
Every engine is independently useful and interoperates through one API-first Trust Fabric — so identity, evidence, policy, risk and certification stay consistent across your AI estate.
Related engines
Incident Management
Contains a misbehaving agent immediately — revoke its mandate, fail closed, and keep the signed record.
- Instant mandate revocation
- Fail-closed containment
- Escalation queues
- Post-incident evidence
Event Fabric
The signal backbone — verification and status webhooks, escalation polling and evidence-flush events.
- Verification / contact webhooks
- Escalation status polling
- Evidence batch flush
- Real-time trust signals
Security Architecture
Security-by-design: signed, fail-closed, PII encrypted at rest, only commitments on-chain, and quantum-resistant channels.
- Ed25519 + hybrid post-quantum
- Encryption at rest (AES-256-GCM)
- On-chain commitments, not PII
- Fail-closed by default
See AI Sandbox in action
Book a demo and see how the platform establishes trust across your AI systems.