Live transcription,
30 languages.
Every word, captured as it’s spoken. Faktum follows the room across languages and speakers without a setup step.
- Real-time, on-device transcription
- A waveform you can scrub afterward

Faktum records and transcribes in 30 languages, then checks each claim against the web in real time — supported, contested, contradicted, or countered.

Faktum listens, isolates each checkable claim, searches the web, and returns a verdict — without anyone lifting a finger.
Records and transcribes the room in 30 languages.
Isolates statements that can actually be checked.
Searches sources and weighs what they say.
Returns one of four verdicts with its sources.
We looked at the numbers again last quarter. , and on retirement, frankly, .
SpeakerOn the policy side, the figure was that . My own view — — is where we should land.
Each underline is a checkable claim. Activate one to see Faktum’s verdict and its sources.
Select a claim from the transcript to see how Faktum weighs it — verdict, confidence, and the sources behind it.
Every word, captured as it’s spoken. Faktum follows the room across languages and speakers without a setup step.

Each claim gets a verdict — supported, contested, contradicted, or counterpoint — and the sources Faktum used to reach it.

When it ends, Faktum writes the overview, the positions taken, the open questions, and the action items.

Memory runs on-device. Raw transcript is never used to train models. Processing is designed for EU data residency.

A recording indicator stays on screen the whole session — on the lock screen too.
Follows the room as it switches.

Memory stays on your device.

From the people who sit in the most claim-heavy rooms.
“It flagged a revenue figure as contested while the analyst was still talking. We paused, checked the source, and corrected the deck before it ever left the room.”
“I run interviews live, and Faktum sits quietly in the corner sorting claims by verdict. By the time I ask my follow-up, the contradiction is already on screen with sources.”
“My students used to argue from memory. Now they review the session afterward and see exactly which claims held up and which fell apart — with the receipts.”
“The summaries separate what was agreed from what is still an open question. That alone saved me a half-hour of writing notes after every policy briefing.”
“On investor calls it is the calmest thing in the room. No alerts, no noise — just a record of every claim and whether the web backs it up.”
One app, both platforms. Upgrade only when your meetings need it.
For trying Faktum in your own meetings.
For people who run meetings every day.
For organizations with shared workspaces.
A fact-checker you don’t have to fact-check. The architecture, stated plainly. No marketing numbers — just how it works.
What Faktum does, where it runs, and how it treats your data.
Faktum is a passive AI meeting assistant for iOS 17 and macOS 14. It records and transcribes meetings in 30 languages and fact-checks each spoken claim against the web in real time, returning one of four verdicts — supported, contested, contradicted, or counterpoint — each with its sources.
Faktum runs on iOS 17 and later and macOS 14 and later. One app covers both platforms, with no separate setup step.
30 languages, auto-detected. Transcription and fact-checking follow the room as speakers switch languages, without a manual switch.
Every checkable claim gets one of four verdicts — supported, contested, contradicted, or counterpoint — tied to the exact quote and citing the sources Faktum used to reach it.
Memory runs on your device and holds only derived insights, never the raw transcript. Audio is analyzed live and discarded once a claim is checked — it is never stored on Faktum's servers. Saved summaries sync through your own iCloud, and processing is designed for EU data residency.
Faktum is free to start (€0): record, transcribe, 30 languages, a limited number of fact-checks per meeting, plus summaries and action items. Pro is €14.99/month or €149.99/year and adds unlimited fact-checks, on-device memory across meetings, and a lock-screen live activity. Team plans are custom with volume pricing.
No. There are no modals, sounds, or paywalls mid-session. A red recording indicator stays visible the entire time Faktum is listening, including on the lock screen.
When a meeting ends, Faktum writes an overview, the positions taken, the open questions, and action items ready to share.
Faktum transcribes and fact-checks 30 languages, auto-detected as the room switches: Chinese, English, Cantonese, Arabic, German, French, Spanish, Portuguese, Indonesian, Italian, Korean, Russian, Thai, Vietnamese, Japanese, Turkish, Hindi, Malay, Dutch, Swedish, Danish, Finnish, Polish, Czech, Filipino, Persian, Greek, Hungarian, Macedonian, and Romanian.
No modals · no sounds · no paywalls mid-session.