Pinecone VDB Provider

The Pinecone VDB Provider is a database module that works together with the AI Core module, to allow you to do vector searches with AI Search module or other modules that implement it.

Overview

This Drupal module provides integration with Pinecone, a managed vector database and works together with the AI Core module. It includes features for inserting, deleting, and managing vector data.

This allows you to use the AI Search features from the AI Core module including its:

• Ability to find highly relevant results from keywords or full sentence queries
• Integration with Views via Search API, including boosting of Search API Database and Search API SOLR setups
• Ability to use as a source for Retrieval Augmented Generation (RAG)

Requirements

• Pinecone serverless account (starter is fine) and API key.

Installation

1. (In Drupal) Enable this module (which requires the AI Core and Search API modules)
2. (In Drupal) Ensure you have an Embedding type configured in the AI Core module configuration at Admin > Configuration > AI.
3. (In Drupal) Configure the API connection to Pine via Admin > Configuration > Vector Database Providers > Pinecone.
4. (In Pinecone) Create an Index in the Pinecone UI via pinecone.io.
5. (In Drupal) Create a new Search API Server, select Pinecone as the backend, and select that Index.
6. (In Drupal) Set up AI Search as desired (see AI Search documentation).

Contributing to the Pinecone PHP library dependency

See https://github.com/probots-io/pinecone-php

Compatibility with Pinecone options

This module is intended to support the 'Serverless' architecture only. The VDB
provider methods around Collections therefore map to Pinecone's namespace
architecture.

If you intend to use Pinecone's Pod infrastructure, that is not yet supported;
however, can be contributed back as a new VDB Provider within this module.