Disruptive RainDisruptive RainCookbook
AboutTopicsGuidesAPI docsSource
All guides

RAG on PDFs with file search

Chunk documents, search with citations, and answer precisely.

Intermediate20 min readMar 11, 2025
ResponsesSearchCitations
Retrieval + memoryText workflows

Actions

Open in GitHubView as Markdown
Key takeaways
  • Use chunk overlap to keep context intact.
  • Return citations for every claim.
  • Score retrieval quality with evals.

Chunking strategy

Split PDFs into overlapping chunks with metadata like page and section labels.

Retrieval and reranking

Retrieve with embeddings, then rerank for precision before answering.

Citations

Attach citations and section IDs to every response so users can verify sources.