Research with your own sources
Most of research is deciding what a source actually says. KnowNote is built around that: one notebook per question, answers that carry the passage behind them, and notes that survive the session.
The shape of the work
Not a feature list — the loop this tool is shaped around.
- Collect
- papers, manuals, saved pagesA notebook is the unit of context. You put the material for one question in one place, and retrieval never ranges outside it.
- Ask
- in your own wordsRetrieval is multilingual, so a question written in one language can surface a passage in another. That matters when the literature is not in the language you work in.
- Verify
- the passage beside itEvery answer carries the passages it was built from, with the offsets they were cut at. The point of asking is to check the text behind an answer, not to take it on faith.
- Keep
- structured notesWhat you concluded goes into a note beside the sources, in the same application, so the next session starts from your thinking rather than from the documents again.
- Revisit
- mind map, notebooksA notebook can be turned into a mind map to see its structure at a glance, which is usually how the gaps in your reading become visible.
Why the sources stay on your machine
A literature review is often about material you cannot upload: an NDA'd specification, a client's contract, patient notes, a pre-release paper. A tool that requires a cloud account to read your own documents is unavailable exactly when the documents matter most. KnowNote has no upload step at all — the embedder runs in the app, and the only outbound request is the one that writes the answer, to an endpoint you chose.
What it will not do for you
Retrieval is a lens on the material you brought. It is not a literature review.
- It does not search the web
- your sources onlyThere is no browsing and no search index behind it. If a fact is not in the documents you imported, KnowNote has nothing to say about it.
- It does not rank by authority
- by similarityRetrieval returns the passages closest to your question, not the most trustworthy ones. A weak source you imported is as retrievable as a strong one.
- It does not read tables and figures
- text extractionImport extracts text. A PDF whose meaning lives in a chart, a scanned image or a complex table will lose that meaning on the way in.