Metalogue
Docs

Pinecone Integration

Connect your Pinecone vector database to Metalogue

Pinecone Integration

Connect your existing Pinecone indexes to federate queries across vector stores.

Quick Start

const connector = await client.createConnector({
  connector_type: 'pinecone',
  display_name: 'Production Pinecone',
  credentials: {
    api_key: 'xxxxxxxxxxxx',
    environment: 'us-east1-gcp',
  },
});

await client.syncConnector(connector.connector_id, true);

Authentication

{
  "credentials": {
    "api_key": "xxxxxxxxxxxx",
    "environment": "us-east1-gcp"
  }
}

Configuration

{
  "settings": {
    "indexes": ["production", "staging"],
    "namespace": "documents",
    "embedding_model": "text-embedding-ada-002",
    "dimension": 1536
  }
}

Vec2Vec Translation

When querying across different embedding models, Metalogue automatically translates embeddings:

// Query returns results from Pinecone even if using different embedding model
const results = await client.query({
  text: 'quarterly revenue projections',
  embedding_model: 'text-embedding-3-small', // Different from Pinecone's model
});

Metadata Filtering

Pass Pinecone metadata filters:

const results = await client.query({
  text: 'product roadmap',
  filters: {
    connector_id: 'pinecone-123',
    metadata: {
      department: 'engineering',
      year: 2026
    }
  }
});

Rate Limits

  • Queries: Based on Pinecone plan
  • Batched operations: Automatic chunking

Next Steps