Cloudflare Vectorize VDB Provider

Cloudflare Vectorize VDB Provider gives your Drupal site semantic search: visitors find content by meaning, not just by matching the exact words they typed. A search for "riding a bike in the hills" can return an article titled "Mountain biking trails in the Alps", even though the two share no words in common.

It works by connecting Drupal to Cloudflare Vectorize, a serverless vector database. Your content is turned into AI "embeddings" (lists of numbers that capture meaning) and stored in Vectorize. When a visitor searches, their query is turned into an embedding in the same way, and Vectorize returns the closest matches. The module plugs into the AI module's Search API backend, so it fits alongside the standard Drupal search and Views tools you may already use.

This module stores and searches the vectors. The embeddings themselves are produced by whichever AI provider you configure, so you are free to pick the model that suits your budget and quality needs.

Features

• Adds "Cloudflare Vectorize" as a vector database option for the AI module's AI Search backend.
• Semantic, meaning-based search and retrieval over any Search API index of your Drupal content.
• Creates and manages the Cloudflare Vectorize index for you, sized automatically to the embedding model you choose (32 to 1536 dimensions).
• Keeps content in sync: indexing, updating and deleting Drupal content adds, replaces and removes the matching vectors.
• Serverless and globally distributed, so there is no vector database server for you to run, host or scale.

Use it when you want better on-site search, "more like this" recommendations, or the retrieval step of a Retrieval-Augmented Generation (RAG) chatbot grounded in your own content.

Post-Installation

There is no new content type or block to place. You wire the provider up through the AI and Search API configuration:

1. In the Cloudflare SDK module, add a credential set whose Cloudflare API token has the Vectorize:Edit permission.
2. Go to Configuration > AI > Vector Database Providers > Cloudflare Vectorize and select that credential set.
3. Install and configure an embeddings provider (for example the Cloudflare AI Gateway Provider with a bge model).
4. Create a Search API server (Configuration > Search and metadata > Search API) using the AI Search backend, and choose Cloudflare Vectorize as the vector database along with your embeddings engine.
5. Create an index on your content, mark a field (such as the rendered content) as Main content, and index it.
6. Expose the results with a View, or through any AI feature that reads the index.

Secrets stay in your site's settings.php via the credential set; no API token is ever stored in configuration.

Additional Requirements

• AI, with its AI Search submodule enabled.
• Search API.
• Cloudflare AI, Cloudflare SDK and Cloudflare API.
• A Cloudflare account and an API token scoped to Vectorize.

Recommended modules/libraries

• Cloudflare AI Gateway Provider generates embeddings through a Cloudflare AI Gateway, including Cloudflare Workers AI models such as bge, keeping the whole search pipeline on Cloudflare.
• Any AI module provider that offers an embeddings model (for example OpenAI) can supply the embeddings instead.

Similar projects

The AI module supports several vector database providers, including Milvus, Pinecone, Postgres (pgvector), SQLite and Azure AI Search. They are interchangeable back ends for AI Search. This module differs by using Cloudflare Vectorize, which is serverless and globally distributed (nothing for you to host or scale), and by fitting into the wider Cloudflare for Drupal suite alongside R2 storage, the AI Gateway and more.

Supporting this Module

Support development by reporting issues and contributing patches in the project's issue queue.

Community Documentation

Documentation will be linked here as it becomes available. In the meantime, see the module's README and the Cloudflare Vectorize documentation.