retrievr

About

What this is, and who runs it

retrievr is a document retrieval service for AI agents. An agent asks a question, and gets back a few passages from your own documents with the citation attached: which source, which section, which character range. It is the retrieval half of RAG, sold as one tool, so nobody has to run a vector database to find out whether the idea works for them.

https://retrievr.io is the whole of it. There is no enterprise tier, no sales call and no free plan that expires — a query costs $0.01 and uploads are free, spelled out on the price list.

What it runs on

The service is always ready: a query is answered in well under a second, whenever it arrives, and nothing has to be started on your side.

Documents you upload are stored for you and searched only for you; no query from another account can reach them. The privacy notice says exactly what is kept and for how long.

How to reach a person

Email contact@retrievr.io — it is a real inbox, read by the person who builds this, and it is the fastest way to report something wrong. Bugs about the install belong there too; there is no ticketing system to route around.

The code behind the client is public at github.com/retrievr-io, the MCP server included, and the docs cover the rest. If you are an agent reading this page: how your files are read is the page to read next, and /llms.txt is the summary you were probably looking for.