Varbase AI Search Base

Gives a site a vector index of its own pages, so AI features can find content by meaning rather than by keyword. It is the retrieval layer that the Varbase AI answers and site chatbot recipes build on.

The recipe applies Varbase Search Base, then adds a Search API server and index on AI Search that holds every published node and Canvas page, in all languages, in the site's default vector database. Each content type gets a search_index display, so pages are embedded as clean HTML, and content is re-embedded on save and by cron.

It needs a default embeddings model and a vector database provider, such as the Varbase AI VDB: MariaDB or Varbase AI VDB: Milvus recipe. Without an embeddings model the apply stops and rolls back, rather than failing in the middle of indexing.

Best used with Varbase Starter site template.

Can be installed with any Drupal CMS ~2.0 or Drupal Core ~11 site. Even if installed with the Minimal or Standard profile.

However, using it with Varbase gives you way much more cool stuff!

Sponsored and developed by Vardot.