AI Recipe: Answers

Generic setup for the ai_answers RAG search-and-cite Q&A feature: a vector search index over all site content, a retrieval agent restricted to that index, and placeable Question/Answer/Sources blocks.

The ai_answers module is what actually answers questions: it takes a visitor's question, retrieves matching content through Search API/AI Search, and returns a cited answer with sources. This recipe exists to give it something real to search: the vector index, agent, and blocks described below.

This recipe installs and wires up Search API, AI Search, AI Agents, and AI Answers: it clones every node type's default view display into a dedicated view mode so indexing can embed the same rendered HTML a visitor would see (no need to know any site's field names), creates a Postgres/pgvector-backed Search API server and chunked-tracker index, creates an AI Agent whose RAG tool is forced to that index, enables AI Answers for it, and places the three AI Answers blocks with no visibility restriction.

It's deliberately unopinionated about what gets indexed, what the agent sounds like, and where the Q&A widget appears - those are exactly the things a real site needs to decide. Extend it with a recipes: [ai_recipe_answers] dependency and override the same config names via setProperties/simpleConfigUpdate actions to tailor the index's field mapping, the agent's instructions, and the blocks' placement/copy.

Requires:

• A default embeddings provider already configured (Configuration > AI > AI settings)
• A reachable Postgres instance with the vector extension (see ai_recipe_vdb_provider_postgres, which this recipe depends on)
• The ai_search RagTool patch (issue #3584031) - without it, AI Answers silently gets zero retrieval results on every question even though search itself works

FYI: if the content you want indexed doesn't already live in Drupal (pages from other websites, for example), crwlr crawls external sites into structured Drupal content through a configurable admin workflow (CSS/XPath field mapping, robots.txt compliance, batch processing). Content it creates can then be indexed by this recipe like anything else.

See the project's README for full apply instructions, known limitations, and testing steps.