AI Search

AI Search integrates vector embeddings and semantic search into Backdrop CMS
via the Search API framework. It provides a provider-agnostic embedding
pipeline, multiple vector storage backends, and optional UI modules for
conversational search and chatbot experiences.

Features

• Semantic search — indexes content as vector embeddings and retrieves
  results by meaning rather than keyword match.
• Multiple embedding providers — OpenAI, Ollama, and OpenRouter embedding
  models supported out of the box.
• Multiple vector backends — Milvus/Zilliz, MariaDB 11.7+ Vector, PostgreSQL with
  pgvector, and Pinecone all supported as Search API service backends.
• Hybrid search — optionally boosts traditional database search results
  with AI similarity scores.
• Text chunking — splits large content into chunks before embedding for
  better retrieval accuracy.
• Extensible — add new embedding engines or vector backends by implementing
  a small set of hooks and base classes.

Requirements

• Backdrop CMS 1.x
• PHP 8.2 or later
• AI module with at least one
  configured provider
• Search API
• At least one vector backend (ai_search_provider_milvus, ai_search_provider_mariadb, ai_search_provider_pgvector, or ai_search_provider_pinecone)

Installation

1. Install and enable the AI module and configure an embedding-capable provider.
2. Install and enable this module and your chosen vector backend module.
3. Configure your vector backend server in Search API (Admin → Configuration → Search and Metadata → Search API).
4. Create a Search API index on that server. Ensure your embedding model dimension matches your vector database index dimension.
5. Add your content fields to the index using the Embeddings field type.
6. Index your content.

Included Submodules

ai_search_search_block

A configurable block that provides AI-powered site search. An optional logging
submodule (ai_search_search_block_log) records queries and results.

ai_search_explorer

Admin UI for running vector search queries and inspecting results interactively.
Found at Admin → Configuration → AI → Search API AI Explorer.

ai_search_simple_chatbot

Block-based chatbot with streaming responses and session-based chat history,
powered by the configured AI search index. (Deprecated in favor of the more flexible and powerful
AI Assistants module.)

Issues

Bugs and feature requests should be reported in the
Issue Queue.

Current Maintainer

Justin Keiser

Credits

• Created for Backdrop CMS by Justin Keiser.
• Inspired by the Drupal Search API AI module and the AI Search
  module.
• Developed with AI assistance.

License

This project is GPL v2 software. See the LICENSE.txt file in this directory for
complete text.