Cryterio's tools

It is Cryterio's core function. While the person speaks, we isolate each factual claim, check it against the sources and publish the verdict on the real-time timeline, with a label, a confidence level and an explanation. Opinions, promises and rhetoric are left out: we only check what can be tested against a fact. The work is done once per broadcast and shared with every viewer.

Every verdict carries the exact minute of the broadcast at which the claim was made. You can jump back to the precise point to hear the context, and anyone reading later understands what the verdict refers to. It's the thread that ties what was said to what we verified.

Every verdict has its own page, stable over time and citable: it holds the original claim, the label, the explanation, the sources with link and extract, and the date. You can share a single verdict in an article, on social media or in a chat, and whoever opens the link sees exactly the same card. A fact-check you can link to, not just watch.

The permalink provides the signature proof (Ed25519 algorithm) and the official public key is on the Method page: anyone can check it themselves, even offline. The signature proves the text is the one we published, not that the verdict is correct.

The most common public broadcasts — rallies, press conferences, debates streamed on YouTube or Vimeo — go straight into the verification schedule: the public stream's URL is enough. We take the audio, transcribe it on our own servers and start checking in real time. It's the main channel, and already fully operational.

No live stream needed: a rally or interview already recorded (audio or video) can be uploaded and run through the same pipeline — transcription, claim isolation, source-anchored checking. You get the same timeline of verdicts, useful for analysing an event or an archive item after the fact.

A lightweight box with a broadcast's verified timeline, to drop into an article or a third-party site with a single line of HTML. The verdicts update live inside the hosting page, without weighing it down. That way you bring verification straight to where your audience already reads.

Every published verdict stays available: a full-text search lets you find a figure, a claim or a topic across past broadcasts. Useful for anyone who works with facts and wants to quickly retrieve what was said and verified, and when.

A broadcast's verdicts export to CSV for the newsroom's internal analysis: claims, labels, confidence, sources and time references in a spreadsheet. So you can work the data with your own tools, build tables or summaries, fold it into a piece.

Verification and interface are available today in Italian, French and English; German and Spanish are coming. It means you can fact-check public speech in several languages and offer the whole experience — site, timeline, verdicts — in your audience's language.