Model Gate
Pre-inference allow or deny with failing rule named.
Open toolAI & enterprise
AI policy gates and agent permissions for enterprise teams
Your agent should not get a vote on spend, scope, or model choice. Policy evaluates first. The model runs second.
A bank's internal copilot proposes tool calls all day. Model Gate evaluates registry, residency, and role rules before inference. Agent Trust covers longer-lived permission receipts.
AI governance enterpriseLLM guardrailsEU AI Act readinessagentic AI securitymodel registry policy
Tools in this cluster
Twenty-three public demos prove the same signed yes or no shape. These are the ones we see ai & enterprise teams open first.
Agent Trust
Know Your Agent permissions and session receipts.
Open toolEdge Audit
subject_ref proofs for RAG sources on the host.
Open toolPQC Gauge
Post-quantum attestation readout for security architecture.
Open toolLive industry proof
Runs a real audit_proof on api.affix-io.com using the kris@affix-io.com edge-audit service key. Attestation is ML-DSA-65.
Click run to prove.
Free Hub trial
100 proofs. No card. Yours in minutes.
Sign up free, confirm your email, and redeem 100 SDK proof credits for 30 days. Run the same benchmarks on your stack before you commit to a licence.
- Free sign-up at Hub
- 100 proofs after email confirmation
- 30-day window to use them
- No overage on trial
AI & enterprise questions
Does Model Gate call the model?
No. It returns allow or deny before the prompt is sent.
Can agents hold long-lived permissions?
Agent Trust models session-scoped receipts your app can verify.
What signature algorithm is used?
ML-DSA-65 on production attest paths.