Skip to content

Primer

Primer answers questions about your own documents. You upload sources into private libraries, and it answers from them alone, with citations back to the passages it used.

It is built for people who cannot or will not send their documents to a hosted service: everything runs on your infrastructure, against your own model endpoints.

What it does

  • Private libraries. A library belongs to one person. Nobody else can read it, list it, or discover that it exists.
  • Real documents. PDF, Word, PowerPoint, Markdown, and plain text, including text that exists only inside images.
  • Answers with citations. Every claim points at a passage in a specific version of a specific document, so you can check it.
  • Your models. Any OpenAI-compatible endpoint, hosted or local.

How it fits together

Primer system architecture Primer system architecture

Four processes, each owning one thing. The Control API owns what exists and who may see it. Ingestion workers turn uploaded files into passages. Retrieval is the only process that touches a vector store. The web app is what you use.

Where to start

This documentation describes what is built

Primer is under active development. Pages document behaviour that exists and is tested, and say so plainly where something is not built yet.