Dragwyb Docs Dragwyb Docs
Visit Website
Updated on Oct 17, 2026

Connect Pinecone Vector Storage

Connect a Pinecone index to Dragwyb Click to Chat with its API key, host, index name, and a matching embedding configuration.

Dragwyb Click to Chat can use Pinecone as a cloud vector database for the knowledge-base embeddings. You need a Pinecone index and its connection details, plus an embedding provider configured for the text being indexed. The plugin’s settings do not create a Pinecone index for you.

Prepare the Pinecone index

  1. Sign in to the Pinecone Console and create or choose an index suitable for your embedding vectors.
  2. Make sure the index dimension matches the selected embedding provider: the current OpenAI text-embedding-3-small configuration uses 1536 dimensions, while Google Gemini gemini-embedding-001 uses 768 dimensions.
  3. Copy an API key from Pinecone and the exact host URL for the index. In the Pinecone Console, open the index to find its host.

Connect Pinecone in Dragwyb

  1. In WordPress, open Click to Chat → AI Assistant → Knowledge Base → Database & Indexing.
  2. Under Database Storage Engine, select Pinecone.
  3. In Pinecone Configuration, enter the Pinecone API key, host URL, and exact index name.
  4. Under Embedding Configuration, choose OpenAI or Google Gemini and verify that its API key is configured in the AI Engine & Prompt settings.
  5. Save the Knowledge Base settings and select Index Content Now to index the configured sources.
  6. Wait for indexing and embedding processing to finish, then ask a test question whose answer is in the selected source content.

Troubleshooting

  • Connection or request errors: Recheck the API key, host URL, and index name for typos and ensure the host belongs to the intended Pinecone index.
  • Dimension mismatch: Check the index dimensions against the selected embedding provider before re-indexing.
  • No relevant results: Confirm the source was included and indexed, then test with a specific question grounded in that source.

Changing vector storage does not itself create source embeddings; indexing still needs to complete. No live Pinecone connection or indexing request was run while preparing this guide. For the local alternative, see Configure SQLite Vector Storage. For the complete retrieval workflow, see Knowledge Base and Retrieval.