Minna

Minna is now accepting Beta Testers!


Minna turns your lectures, assignments and papers into a knowledge base you can search in natural language. Minna is a completely on-device research assistant. Your papers, results, assignments and lectures never have to leave your computer.

Private Beta

Minna is currently in private beta! You can get access to Minna by filling out the following form. Your email address and name are used to send you a beta invite using Apple's TestFlight app.

All your content, in one place

You can drop an entire semester's worth of course resources into Minna. Organizing that material is simple, and allows you to group items into folders & subfolders. Every file is taggable with a color so you can quickly identify groups of concepts.

The Minna library view showing a Research Methods folder as a grid of color-coded PDF cards — lecture slides, papers and course notes — with a sidebar tree of nested knowledge base folders.

Instant Natural Language Search

Minna leverages on-device machine learning to search thousands of documents faster than the blink of an eye (~12 ms per search). No matter where a concept is, Minna can find it. Search by concept, specific phrases, or authors and Minna will find the content you are looking for.

Chat across your entire database

LLMs are a great tool for research, however their output can't always be trusted. Minna requires that an LLM only makes claims that are backed up by real results. Before answering, the LLMs must search your knowledge database to find answers. Once it finds the answers it needs, it will tell you directly where they are.

The Minna chat view answering a question about variable renaming. The assistant searches the knowledge base, then gives a cited answer with numbered footnotes linking to a References sidebar of source excerpts from the matching paper.

Dig into a specific document

Sometimes you don't want results from your entire database. When that is the case, you can open a specific document and ask questions. In this chat mode, Minna will search directly from the document you are in, and will not go outside of it for results.

A single-document view in Minna: a highlighted, hand-annotated neuroscience paper on the left, and a chat panel on the right answering a question about the paper using only excerpts from that document.