Stop cost spikes
Caps, budgets, alerts
Enforce project budgets and token caps before provider spend runs away.
ENTERPRISE LLM GATEWAY
The control plane for LLM traffic.
OpenAI-compatible gateway for budgets, PII guardrails, audit logs, and reliability routing.
POLICY • AUDIT • BUDGETS • ROUTING
OpenAI-compatible Audit-first controls
Cloudflare-native edge
Policy decisions • Request trace timeline • Budget actions
Sentinel Primo Console
PII detected -> blockedPrevent leaks + PII
Redact / block before storage
Apply redact or block decisions at ingress so sensitive data does not leave your boundary.
Stay up during provider flaps
Retry + fallback routing
Fail over cleanly with retry and fallback policies when primary routes degrade.
Who this is for
Built for platform, security, and FinOps ownership
Platform teams
Standardize LLM ingress without SDK rewrites across services.
Security teams
Apply policy controls and preserve auditable request evidence.
FinOps teams
Enforce budgets and stop runaway token spend before invoices.
Core capabilities
Signature control-plane bento
Budgets & caps
Set monthly and per-project spend caps with deterministic enforcement actions.
Budget view
Threshold alert at 80%, hard block at 100%.
Rate limits (TPM/RPM)
Control throughput by tenant, project, and key profile.
PII detect + redact
Evaluate content for sensitive entities and redact or block automatically.
Key hygiene
Keep provider keys isolated and rotated with scoped project controls.
Audit logs + retention
Export structured audit records on configurable retention profiles.
Routing + fallback
Map logical models to multiple providers with retry and failover.
Analytics
Track latency, tokens, and cost by environment and team.
Caching controls
Reduce duplicate spend with policy-aware response caching.
How it works
Three steps to enforce policy at the edge
Step 1
Swap base URL
Keep your existing OpenAI SDK flow and point requests to Sentinel Primo.
Step 2
Apply policies
Run allow, block, and redact checks before provider execution.
Step 3
Observe & enforce
Review audit logs, budget actions, route decisions, and reliability outcomes.
Routing diagram
Control-plane proof
Internal product signals, not marketing noise
Requests view
Live request queue with policy and route outcomes.
PII redactedFallback triggeredToken usage logged
Policy decisions
Deterministic policy path with redact and block evidence.
Allow / Block / RedactRule match traceProvider call gated
Budgets and caps
Budget governance with threshold and enforcement signals.
Budget exceeded: blockedSoft threshold alertsPer-project caps
Security & trust
Controls built for high-trust operations
Data handling
- Logs store operational metadata, not full payloads by default.
- PII detection can redact or block before provider calls.
- Provider keys remain scoped per project and environment.
Retention & audit
- Retention policies are configurable by workspace or environment.
- Audit events are structured for review and export.
- Cost, policy, and routing outcomes stay traceable per request.
Isolation & access
- Project-level API keys support workload separation.
- Workspace boundaries prevent cross-team spillover.
- SSO and advanced RBAC are clearly tracked as roadmap items.
FAQ
Do you log raw prompts and raw model responses? +
Not by default. Sentinel Primo is designed around metadata-first logging and configurable retention controls.
How does PII handling work? +
Policies can detect sensitive entities and apply allow, redact, or block actions before provider forwarding.
Can we configure retention windows? +
Yes. Retention profiles are configurable so teams can align operational visibility with internal policy.
Are provider keys exposed to application clients? +
No. Provider credentials stay server-side inside Sentinel Primo policy and routing layers.
Can we export audit records? +
Yes. Structured audit exports are available for compliance, finance, and security workflows.
Resources
Operational guides for rollout teams
Quickstart (5 min)
Swap your base URL, verify request flow, and enable baseline policy checks.
Cost control (budgets)
Define spend guardrails and enforce deterministic budget actions.
Reliability
Design retry and fallback behavior for provider outages and latency spikes.
Security & privacy
Configure PII treatment, metadata logs, and retention posture.
Early-access pricing is available for design partners. Request a demo to discuss fit.
Talk to salesBring governance to every model request
Review architecture fit, policy posture, and rollout sequencing with the Sentinel Primo team.