Use case

Meeting minutes with the screens of what was shown

The meeting was recorded, somebody shared their screen, and what is left is an 800 MB file nobody is going to reopen. The minutes become a document with the moments that matter, the screen of each one and the speech paired to it.

Turn a meeting into minutes use the recording you already have

The pain

Whoever writes the minutes rewatches the meeting. An hour of meeting costs an hour of someone, plus the time spent going back three times to the bit where the number was said. And when the decision is challenged two weeks later, the answer is at minute 47 of a file nobody wants to open.

A plain transcript solves half of it: it says what was said, not what was on screen. In a project meeting, half the decisions happen while looking at a number in a shared report — and that number is not in the transcript.

What comes out

Minutes in a compact grid: the screens in sequence, and the full transcript at the end for anyone who wants to check.

• Only the screens that changed — an hour of meeting with one static slide does not become sixty identical images
• Every moment with its clock time, anchored to the real start time of the meeting
• The two audio channels kept apart: what came from the microphone and what came from the computer are labelled in the transcript
• Meeting header: subject, place or platform, who chaired it and who attended
• A prompt ready for the AI to write the minutes — topics, decisions, open items with owners and next steps, with the timestamp of each decision
• The transcript is yours: export it as .vtt, .srt or .txt, or translate it into another language without the text leaving the device

How to do it

1. Pick the Meeting minutes use case in step 1.
2. Drag in the meeting recording — Teams, Meet, Zoom, wherever. The file is read by the browser and does not leave your computer.
3. Let it transcribe and pull the frames. If you already have the caption file from Drive or Teams, paste it in place of the transcript.
4. Review, produce the document, and paste the prompt into the AI your company already approved — together with the PDF.

What it does not do — better to know up front

It does not write the minutes for you. It produces the document with the screens, the speech and the timestamps, plus the prompt ready for the AI you use. Embedding a language model in the browser today would be weight with no gain — and sending your meeting to a server of ours would break the only promise this product makes.

It does not name who spoke. It separates the two audio channels (microphone and computer), which is what can be done honestly without voice recognition. On a call with eight people, every remote voice arrives on the same channel.

It does not join the meeting as a participant. There is no bot of ours on your call, no invitation to accept, and no recording hosted anywhere.

Turn a meeting into minutes See the plans

It works for project meetings, steering committees, agile ceremonies and system demos — anything where the screen is part of what was decided.