Launching August 2026

Put every AI request under policy.

Cardean Gateway is a Rust-native control plane between your applications and AI providers. Use one OpenAI-compatible interface to route traffic, protect sensitive data, enforce policy, control usage and retain a useful audit trail.

Join the Gateway launch Talk to an engineer

Runs self-managed or in controlled cloud environments.

A beam of AI request traffic passing through a governed gate before reaching model providers
Integration

Keep the interfaces your teams already use.

Works with applications and tools built around OpenAI-compatible APIs, including common Python and TypeScript SDK patterns.

Core outcomes

What Cardean Gateway delivers.

Govern access

Decide which teams, applications and environments may call which models, under which limits and with which credentials.

Protect data in motion

Inspect and transform requests before they leave your control boundary. Apply rules for sensitive fields, prohibited content and approved destinations.

Keep AI services available

Use routing, timeouts, retries, fallbacks and health signals to reduce the impact of provider or model failure.

Understand usage and cost

Attribute requests, tokens, latency and spend to the projects, teams and services that generated them.

Retain meaningful evidence

Record policy decisions, configuration changes and request metadata in a form that supports operational review and investigation.

How it works

One interface for applications. One policy path for the organisation.

Cardean Gateway request flow Application requests pass through the Cardean gateway, which identifies the caller, evaluates policy, routes the request and records evidence, before reaching an approved AI provider. Requests to unapproved destinations are blocked. Your app OpenAI-compatible CARDEAN GATEWAY 1. Identify 2. Evaluate policy 3. Route 4. Observe Evidence recorded · Logging configurable Approved provider routed Unapproved blocked by policy
Every request: identified, policy-checked, routed and recorded — through one governed path.
  1. Connect applications to a Cardean endpoint using a familiar API shape.
  2. Identify the calling project, service, team or environment.
  3. Evaluate the request against access, data, model, rate and budget policy.
  4. Route approved traffic to the appropriate provider or deployment.
  5. Observe the response, latency, usage, policy outcome and relevant metadata.
  6. Record the evidence needed for operations, cost control and review.
Deployment

Run the control plane where your requirements demand.

Cardean Cloud

Managed by Cardean for teams that want the fastest operational path.

Private environment

Deploy into a customer-controlled cloud account or isolated environment with Cardean support.

Available for selected design partners.

Self-managed

Operate Cardean Gateway on infrastructure your team controls.

Who it is for

Built for the teams accountable for AI in production.

Platform engineering

Offer an internal AI service with reusable controls.

Security and privacy

Put enforceable decisions around data movement, provider access and logging.

AI governance

Connect approved-use policy to actual model traffic and evidence.

Product teams

Keep a familiar interface and avoid rebuilding provider integrations and controls.

Quickstart

Change the endpoint, then add policy deliberately.

Illustrative request — the documentation has the full interface.

illustrative-request.sh 
# Illustrative request — endpoint and model alias shown are examples only.
curl https://gateway.example/v1/chat/completions \
  -H "Authorization: Bearer $CARDEAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "approved-model-alias",
    "messages": [
      {"role": "user", "content": "Summarise this case note."}
    ]
  }'