How I find an answer
Scroll to replay one real question: “What are the exclusion criteria, and which samples showed strong antigen activation?”
Recorded from the live system on 2 October 2026. Every number below comes from that run.
01Pick the topic
Your question sits much closer to the medical files (0.65) than to finance (0.04), so I only search medical. Deciding took 4.9s and no AI call: it's a similarity score.
02Plan the search
Your question asks 2 things, so I split it: "exclusion criteria for the study" and "samples that showed strong antigen activation".
03Find the passages
For each search I look two ways at once: by meaning, and by exact keywords. Then I merge both rankings, so passages that do well in both rise to the top. I kept the best 8; a bigger star ranked higher.
04Write the answer
I answer only from those passages and cite each one I use. The pink lines join the answer to the 3 passages it cited.
05Check my work
A model from a different AI company checks every claim against the passages. Here 7 of 12 claims were supported (58%). I show this even when it's low.
Lilac: medical passages. Mint: finance. Pink: your question and what it cited.
What I can't do (yet)
"Which is the highest?" questions
I find the closest passages, not every row, so I can miss the true top value.
Small test set
Each topic was tested on 11 answerable questions, so one question moves a score by about 0.09.
Only finance and medical
Files on other topics are refused before any upload is used.
Spreadsheets up to 25 rows
Each row costs one AI call, and the daily budget is shared.
Uploads are temporary
Your files disappear after an hour, or when the free server goes to sleep.
Not advice
Answers are for exploring documents, not for medical or financial decisions.