Tutorial · 5 min

How to use Project Memory (RAG)

A private vector store over your own library

Index your papers and notes into a persistent project memory so every service answers from your own corpus instead of starting cold each time.

Before you start

  • • Documents or a library to index

Steps

  1. 1

    Open Project Memory (RAG)

    Sidebar → Project Memory (RAG).

  2. 2

    Create a project

    One memory project per paper or review keeps retrieval sharp.

  3. 3

    Index your sources

    Documents are chunked and embedded so later questions retrieve the exact passages instead of the whole file.

  4. 4

    Ask across the project

    Answers cite the chunk they came from, so you can trace anything back to the source document.

Tips

  • • Re-index after you add a large batch of new papers.

Troubleshooting

Retrieval misses an obvious passage

Check the document actually indexed — image-only PDFs need OCR before embedding.

More tutorials

Browse all 80 tutorials →