From Zero to Secured Retrieval in 5 Minutes

Step 1: Install the SDK. Run pip install gateco, then authenticate: gateco login --email you@company.com --password secret. This stores credentials locally and you're ready to go.

Step 2: Connect your vector database. Use the SDK or CLI to create a connector — Gateco supports pgvector, Pinecone, Qdrant, Weaviate, Milvus, Chroma, OpenSearch, Supabase, and Neon. For example: client.connectors.create(name="My Pinecone", type="pinecone", config={"api_key": "pk-...", "index_name": "knowledge"}). Then test it: client.connectors.test(connector.id) to verify connectivity.

Step 3: Define a policy. Create an RBAC policy that allows specific roles to access specific data classifications. For example, allow the "engineering" role to access "internal" classified documents. Activate the policy to start enforcement.

Step 4: Bind metadata to your resources. Use the classification suggestion engine (client.connectors.suggest_classifications) to scan your resources and get automatic suggestions, or bind metadata manually via the API. This tells Gateco what classification and sensitivity level each resource has.

Step 5: Execute a secured retrieval. Call client.retrievals.execute(connector_id=..., principal_id=..., query_vector=..., top_k=10). Gateco evaluates all active policies, filters results to only what the principal is authorized to see, and returns the results. The audit log records the full decision trace.

That's it. Your AI agent now only sees what it's supposed to see, with every retrieval logged for compliance.