Papers, answers and searches are kept per project.
Plain terms or boolean. Leave empty to skip searching and only ask about papers you already have.
Answered only from papers already in the library — leave empty on a search-only run.
Older papers are left out of the search.
Nothing retrieved yet
New papers since last time, the project library, papers per year, and the cited answer land here.
Retrieval, then reading
- Search — your terms go to PubMed with the year floor applied.
- Collect — matches join the project library; anything already held is marked as such, so the count of new papers is real.
- Answer — your question is answered from passages retrieved out of that library only, with the title and year beside each claim.
Questions
Can it answer from papers it has not collected?
No. If the library holds nothing relevant, the answer says so rather than filling the gap from memory.
What does a project change?
Papers, answers and search history are stored under the project name. A different name starts a separate library.
Why run a search with no question?
To build the library first. Once papers are collected, run again with a question and the same project name.
Does re-running the same terms duplicate papers?
No. Repeats are counted as already collected, so the new-since-last-time figure stays honest.