Research tool
Dossier answers questions about things you have not read yet, and shows you exactly where each answer came from.
Two ways in. Bring a document and ask it questions. Or name a company you know nothing about and let a research agent go and find the sources itself, while you watch it work. Either way you get the passage, the page, or the link behind every claim, so you can check it before you repeat it.
Sources you bring
Documents
Drop in a PDF, paste text, or point it at a URL. It splits the text, indexes it, and answers questions against it. Answers arrive with the source passages attached, and it says so when the answer is not in the document.
Open documents
Sources it finds
Companies
Type a company name and what you want from them. An agent searches, reads their site, looks for recent signals and names, then hands back a brief with talking points and an opening line. Every tool call it makes is printed as it happens.
Open companies
Why these live together
Both halves do the same job. They take a pile of unstructured text nobody has time to read and turn it into an answer you can defend. The only difference is who goes and gets the text. They share the same page reader, the same streaming protocol, and the same rule: if it cannot show you the source, it does not get to make the claim.
- Nothing is hidden
- Ingest prints each step as it runs. The agent prints every search and every page it opens. You see the work, not a spinner.
- Answers carry sources
- Document answers cite the passages they used and let you expand them inline. Briefs name where each fact came from.
- It admits the gap
- If a document does not contain the answer, it says so instead of inventing one. If research turns up no names, the brief says no names were found.