peekseed / search & research

peekseed · local research workspace

What do you want to understand?

Search across your own PDFs, then ask a model to read the matching sections and answer with numbered citations.

/ search ⌘K history

Start with a question

Add a model link in Settings, then upload a PDF in Datasets.

How this works

Three steps, all under your data root.

  1. Add a link. Point peekseed at an embedding model, and optionally a reranker and an answer model.
  2. Add a dataset. Upload one or more PDFs in Datasets, then build a vector index with your embedding link.
  3. Search and ask. Every result links to its PDF page; send the sections you pick to an answer model and cite each as [n].
peekseed