Question 1 of 13
Walk me through how you would handle a large PDF policy manual so an Agentforce agent receives the most relevant passages when searching.
Strong Interview Answer
I would use a file-based Agentforce Data Library that chunks the PDF into coherent passages, creates a search index, and provides a retriever for runtime retrieval. The retriever should return the passages most relevant to the query instead of sending the entire document to the model. I would test chunk boundaries and representative searches, then refine the retrieval configuration if results are too broad or omit important policy context.
What to Listen For
- Chunking strategy
- Search-index configuration
- Retriever configuration
- Representative retrieval testing
Caution
Duplicating or sending the entire manual does not solve the need for focused passage retrieval.