Hyperresearch

The most troublesome thing about in-depth research tools is the citation: it looks like there is a source, but when you click on it, it doesn’t match. Hyperresearch focuses on this matter. Every citation is checked against the original text and a judgment is given. The numbers in the text are also compared with the source.

In addition to reading the full text of the web page, each run also directly accesses more than 40 data interfaces, including U.S. Census, Labor Statistics, SEC EDGAR financial reports, OpenAlex and PubMed papers. The sources you have read and the corresponding original sentences are stored in your personal database. When you ask questions, you will read them here first. You can also export them to Markdown or let your own agent search.

The project is open source on GitHub, under the MIT license, and the first in-depth report is free. The data interface is biased toward the official US source, so the advantage will be smaller if you are doing domestic market research.