Governance
Budgets and execution caps
Set project and environment guardrails that warn, throttle, or block before spend turns into a reporting problem.
Sentinel AI Gateway
The current flagship product in the Sentinel line
Sentinel AI Gateway is the current flagship product in the Sentinel line from Caldorus. It gives platform, engineering, and governance teams stronger control over policy, routing, telemetry, and operational visibility across production AI traffic.
Sentinel product linePolicy and routingProduction AI operations
Product summary
Built inside the Caldorus parent brand and the Sentinel product line, Sentinel AI Gateway is the current live product for governed AI operations.
- Parent brand
- Caldorus
- Product line
- Sentinel
- Current flagship
- Sentinel AI Gateway
Capabilities
What Sentinel AI Gateway includes
Sentinel AI Gateway is the current flagship product in the Sentinel line. These are its current operating capabilities.
Security
PII detection, redaction, and block policy
Evaluate sensitive traffic before provider forwarding and make privacy controls explicit, reviewable, and auditable.
Reliability
Routing, retries, and fallback policy
Define primary and fallback paths per workload so resiliency follows policy instead of runtime improvisation.
Audit
Metadata-first observability
Capture route decisions, policy outcomes, token usage, and latency signals without defaulting to raw payload retention.
Access
Scoped keys and workload isolation
Separate projects, environments, and teams with clearer control boundaries across the AI request path.
Operations
Analytics for cost and stability
Track request outcomes across policy, spend, and routing so teams can tune controls with operational evidence.
Architecture
How Sentinel AI Gateway works
Within the Sentinel line, AI Gateway is the current product shown in detail. The request path is where it creates control.
01
Ingress without client rewrites
Teams start by pointing existing OpenAI-compatible traffic through Sentinel AI Gateway while preserving familiar request contracts.
02
Policy before provider execution
Requests are evaluated for budgets, data handling, and governance rules before leaving the controlled boundary.
03
Routing with explicit intent
Primary and fallback providers are selected through policy so reliability tradeoffs stay deliberate and observable.
04
Telemetry for review and enforcement
Route decisions, latency, token usage, and policy outcomes stay connected as one operational record.
Operational proof
Current proof points from Sentinel AI Gateway
These screenshots and signals show Sentinel AI Gateway in use today across routing, policy, budgets, and operational review.
Request path visibility
Request traceRoute decisionUsage logged
Policy decision evidence
Rule match tracePII controlsExecution gated
Budget and cap controls
Threshold alertsBlock conditionsPer-project limits
Data handling
- Operational metadata is preserved without default full-payload storage.
- Sensitive entities can be redacted or blocked before provider calls.
- Provider credentials remain scoped and server-side.
Retention and audit
- Retention policy can align to workspace, environment, and review needs.
- Audit evidence stays structured for security, finance, and platform workflows.
- Route and policy outcomes remain traceable per request.
Isolation and access
- Project-level keys support workload separation.
- Workspace boundaries reduce cross-team spillover.
- Future controls such as SSO and finer RBAC fit cleanly into the same model.
Resources
Rollout guidance for Sentinel AI Gateway
Use the current resource set to review quickstart rollout, security posture, spend control, and reliability design for Sentinel AI Gateway.
Getting started6 min read
Quickstart path
Start with compatibility, first-request routing, and baseline policy visibility before moving into stricter controls.
Outcome: Route your first request and confirm policy and audit visibility.
Security7 min read
Security and privacy
Define metadata-first logging, PII handling, retention posture, and explicit audit boundaries.
Outcome: Apply guardrails before requests leave your boundary.
FinOps8 min read
Cost control
Turn budgets into deterministic guardrails and align technical routing behavior with financial intent.
Outcome: Prevent runaway spend without slowing engineering teams.
Reliability7 min read
Reliability
Design fallback, retry, and timeout policy for provider turbulence without losing observability.
Outcome: Protect uptime and keep failover behavior explicit.
Sentinel AI Gateway
Start the product evaluation
Sentinel AI Gateway is available today as the current flagship product in the Sentinel line. Use the shared Caldorus contact path to discuss rollout, pricing, documentation, or security requirements.