Connect Pinecone to Gateco

Pinecone is a fully managed vector database purpose-built for semantic search. Gateco connects to Pinecone as a Tier 1 connector, supporting full document ingestion, retroactive registration, and RAG access control — adding identity-aware retrieval and vector database RBAC on top of your Pinecone indexes. Gateco's policy layer complements Pinecone metadata filtering and namespace isolation by enforcing identity-based access at the retrieval layer.

Step 1 — Create a Pinecone index

If you already have an index, skip to Step 2.

1. Log in to app.pinecone.io.
2. Click Create index.
3. Name: any name meaningful to your project (e.g. production-docs).
4. Dimensions: must match your embedding model (e.g. 1536 for text-embedding-3-small).
5. Metric: cosine (recommended for semantic search).
6. Select Serverless (default) or pod-based depending on your plan.
7. Click Create index.

Step 2 — Get your API key

1. In the Pinecone console, navigate to API Keys.
2. Copy your default API key, or click Create API key to create one dedicated to Gateco.

Step 3 — Note your index host

1. In the Pinecone console, click into your index.
2. On the index detail page, copy the Host URL (e.g. production-docs-abc123.svc.us-east-1-aws.pinecone.io).

Step 4 — Add the connector in Gateco

1. Navigate to Connectors → Add connector → Pinecone.
2. Enter your API key and index name.
3. Click Test connection. Gateco will display the index stats (dimension, vector count).
4. Click Save.

Ingest and retrieve

from gateco_sdk import GatecoClient

client = GatecoClient(api_key="gck_live_...")

# Ingest a document
client.ingest.document(
    connector_id="your-connector-id",
    resource_id="doc-001",
    content="Quarterly revenue report Q4 2025",
    embedding=[0.1, 0.2, ...],
    metadata={"classification": "confidential", "department": "finance"},
)

# Retrieve with policy enforcement
result = client.retrievals.execute(
    query="What was Q4 revenue?",
    principal_id="user-alice",
    connector_id="your-connector-id",
    search_mode="vector",
)
print(f"Allowed: {len(result.allowed_chunks)}, Denied: {result.denied_count}")

Troubleshooting