Deban sits beside you in Google Meet, Zoom or Teams. It listens locally, recognises who is talking, and taps you on the shoulder the moment a question is really yours — with a suggested answer in the meeting's language.
Note-takers tell you what happened. Deban tells you, in the moment, that it is your turn — and why.
Local voiceprints recognise your colleagues across meetings. Anonymous voices get stable labels you can name later.
Your role, your areas, your past promises and TODOs form a memory. A question about "the 3D data" is matched to you, not to the whole room.
A red card for a direct question, a yellow one for "this might concern you", a blue one when you could add something. You decide.
Suggested replies in Japanese or English with three stances — positive, neutral, cautious — and the sources they lean on.
Live subtitles in your language, answers in theirs. Made first for Chinese speakers in Japanese meetings.
Speech recognition and speaker recognition run on your Mac. Only the text sent to the language model leaves it, and never your audio.
Every meeting becomes a searchable record: transcript with speakers, topics, minutes, decisions, TODOs with owners, and the moments Deban nudged you — each with a 👍/👎 that tunes future alerts.
Deban