Nothing here is a quiz and nothing is required. If you would rather just start, the button at the bottom of this page takes you straight to the search box and you will never be sent here again.
Ask the way you would ask a person
No syntax, no operators. Write the claim as you heard it — “my uncle says microwaves cause cancer” works fine. Epistry finds the testable claim inside it and shows you what it is about to check before it runs, so you are never surprised by what it analysed.
The more specific, the better the answer. “Is seed oil bad for you?” is a topic; “seed oils raise inflammatory markers in healthy adults” is a claim evidence can settle. A question with no testable claim in it gets a survey of the landscape instead of a verdict — which is the honest answer to that kind of question.
Claim to fact-check
A yes / no question
Open research question
Click any of those to start with it.
What comes back, and what to look at first
A report is long because it shows its work. Four parts answer the question you came with: the verdict and how confident it is; the competing case, and whether it held up; where the verdict is weakest, stated in one line rather than buried; and every source, with what it was worth and why.
Two things surprise people. “Not enough evidence” is a real answer, not a failure — and it never means the claim is false. And the verdict is computed, not written: language models read and code the evidence, then arithmetic decides. That is why the same question asked twice gives the same answer.
What everything in the menu is
Six destinations. You only need the first one to get value out of this.
- Where you bring a claim. Every analysis starts here, and it confirms the wording with you before spending anything.
- Everything you have run, with its verdict. Analyses are versioned — re-running one adds a version rather than overwriting what it said before.
- Claims other people have already had analysed, free to read. Worth a look before spending a token on something already answered.
- Current stories broken into the specific claims inside them, so you can check the claim rather than the headline.
- Individual studies, assessed on their own terms — how it was designed, how it was run, and whether anything corroborates it.
- The field guide: how to read a paper, spot a fallacy, judge a source. None of it needs us — that is rather the point.
Some of those may not be visible yet — a couple of surfaces are still being switched on, and the menu only shows what is live.
What it costs
One analysis is one token, and the free plan gives you five a month. Reading anything already analysed — anything on Explore, or a shared link — is free, always. If a run fails badly enough that you did not get a real verdict, the token comes back automatically.
Bring something you are unsure about
A claim you have argued about, or one you half-believe. Those are the ones worth checking.
And when you want the long answer to “but how does it actually decide?” — . It is the part we would want to read first about someone else’s tool.
You can come back to this page any time from the account menu.