Research Engine
Type a topic or a question. It goes and finds the evidence, shows you every passage with a link to the page it came from, and only then — if you ask for it — an AI reads that evidence and tells you in plain English what the sources say. Ask something that can be answered yes or no and you get that too.
This finds what the sources say. It does not check whether they are right — that is the fact-checker’s job.
Ask your question
You get 3 free uses a month. The 3 are shared across the whole site — the transcriber, the Content Line and this Research Engine all come out of the same 3. A subscription lifts the limit.
Add pages you already trust (optional)
Nothing is judged yet. You see what it found first.
Where it looks — and what it cannot reach
It searches the whole open internet. Any page a program can open without paying and without signing in: news stories, reports, company filings, studies, forum posts, blogs. There is no approved-sites list.
On top of that it goes to up to 24 official databases — SEC filings, PubMed, the Federal Register, campaign-finance records and the rest. Which ones it uses is chosen from your question, so most questions call for a few and none call for all 24. Your result names the ones it tried and the ones that had nothing to add. That is as well as the web search, not instead of it.
Three kinds of page are shut to us
- Paywalled pages. The site asks for money before it shows the words.
- Login-only pages. The site asks you to sign in first. Reddit does this to anything that is not a signed-in browser.
- Sites whose own rules say programs stay out. X is the one that matters today. We leave it alone.
When a page is shut to us we name it in the result and say why. We never quietly skip it and let you think we looked.
You can open those pages yourself
If you already have an account for one of them — a newspaper you pay for, a paid newsletter, Reddit, a journal through your library — you can let the engine use it.
Install the Research Unlock add-on in your browser. When a page is shut out, press Unlock these with my accounts. The add-on opens that page in your browser, where you are already signed in, copies all of that page's readable text, including any parts the page keeps hidden from view, and sends back only that text. Those passages join the same result.
We never take your password. You do not type it here. It is never sent to us and never stored by us. Your signed-in session stays on your own computer. The add-on answers the research page and nothing else — no other page, on our site or anywhere else, can ask it for anything — and it opens only the addresses that page hands it for this question.
It works best alongside the video research
There is a matching step that goes and pulls what people actually said on camera — the practitioner, the engineer, the person who was there — out of video transcripts. This step brings the written record: filings, studies, reports, news.
Run both on the same topic and you get both halves — what people say, and what is on paper — and you can see where the two disagree.
Your signed-in sources
Sign-in happens on the source's own website, in its own browser profile. This engine never receives a password, a cookie, or a session — only a connection ticket it can revoke.
Checking…
What it found
Nothing yet. Ask a question above.
The answer
Find the evidence first, then ask for the answer.