Silent notetaker on an internal call
Invite the employee to sit in on a team call in silent mode. It transcribes the whole conversation, saves it to Drive, and writes a summary afterward, so nobody has to take notes by hand.
Paste a Google Meet, Zoom, or Microsoft Teams link and your AI employee joins the call as a real participant. It listens the whole time, speaks up when you address it by name, and writes a summary to memory and Drive once the call ends.
Paste a Google Meet, Zoom, or Microsoft Teams link and your AI employee joins the call as a real participant. It listens the whole time, speaks up when you address it by name, and writes a summary to memory and Drive once the call ends. One backend covers all three platforms, so you never pick a different setup for a Zoom call versus a Meet call.
The employee joins in one of two modes. Silent mode is the default: it sits in the call, transcribes everything said, and never speaks. Active mode lets it talk back, but only when you address it directly by name, the same wake-word pattern used on the Voice channel. If it hears no one address it, it stays quiet and keeps listening, so a silent notetaker never accidentally interrupts a meeting it was only invited to watch.
What separates this from a generic recording bot is that it is one of your named employees, not a faceless notetaker widget. It shows up under its own name and avatar, it carries whatever tools, memory, and permissions that employee already has, and the notes it takes land in the same Drive and journal the employee uses for every other channel. A recording tool hands you a transcript; this employee can be asked about the call afterward in the same thread it wrote its notes in.
Meeting minutes bill in real time, about 8.3 credits per minute the employee spends in the call, charged every 60 seconds rather than as one lump sum at the end. That means you see the cost accrue as the call runs, and if your balance runs out mid-call, the meter ends the meeting instead of letting it continue for free. If the employee is speaking when the balance runs out, it finishes its current sentence before leaving rather than cutting off mid-word.
Anything the employee says or thinks through during the call, in active mode, is billed the same way as any other chat message: through the normal token-credit pipeline, not the per-minute meeting rate. The two charges are separate and never double-counted.
Four timers protect you from a bot burning credits in an empty or stuck room. If everyone else leaves and the bot is the only one left, it exits after 30 seconds. If ten minutes pass with no audio at all, it leaves on silence. If it sits in a waiting room without being admitted, or a host never starts the meeting, it gives up after ten minutes on either. A separate 15-minute backstop clears a session that never confirms it joined at all, and a 4-hour cap force-ends any call that runs unusually long, in case the normal end signal is ever missed. Whichever timer fires, the employee can tell you plainly why it left rather than leaving you guessing.
The moment the call is over, the full transcript is written to Drive as a dated markdown file, so it is there to read or search even if the employee never says anything about it. Separately, the employee gets a short private nudge in its own thread, reads the transcript, writes a summary into its memory, and can recap the call back to you without you having to ask. Only that short summary, not the raw transcript, is written into the employee's longer-term memory graph, which keeps a single meeting from flooding its memory with page after page of transcript text.
The employee joins as a voice participant, answers from your knowledge, and turns commitments into tasks.
In the meeting, the employee listens and speaks like a participant rather than a silent recorder. Ask a question out loud and it answers in real time from your training data, documents, and past conversations, so decisions can be made in the room instead of deferred.
Work assigned during the call is captured as real tasks on the employee's board, and a transcript and summary are filed automatically when the meeting ends. The attendee that contributes is also the one that follows up.
Invite the employee to sit in on a team call in silent mode. It transcribes the whole conversation, saves it to Drive, and writes a summary afterward, so nobody has to take notes by hand.
Put the employee in active mode and address it by name mid-call to pull up an answer, a number, or a status update on the spot, the same way you would ask a colleague sitting in the room.
Paste a meeting URL into any chat with the employee instead of opening a separate invite screen, and it joins that call directly from the conversation you were already in.
Because the meeting summary lands in the employee's memory, you can ask it what was decided or discussed in a past call from a normal chat message, well after the meeting itself has ended.
| Before | After |
|---|---|
| Meeting AI listens silently and reports later. | The employee participates in the conversation live. |
| Unanswered questions become follow-ups. | Answers from your knowledge arrive in the room. |
| Action items depend on the note taker. | Commitments become tasks the employee executes. |
| A bot that only records the meeting. | An attendee that contributes and follows up. |
No. It joins silent by default, listening and transcribing only. It only speaks in active mode, and even then only when you address it by name; otherwise it stays quiet and keeps taking notes.
The per-minute meter checks your balance as it bills, and once it runs out the employee leaves the call. If it was mid-sentence in active mode, it finishes that sentence before exiting rather than cutting off mid-word.
It will not wait forever. A bot stuck in a waiting room, or waiting on a host who never starts the meeting, gives up and leaves after ten minutes, and tells you which of the two happened.
The full transcript is saved to Drive as a dated file automatically, whether or not the employee says anything about it. The employee also writes a short summary into its own memory and can recap the call to you in chat without being asked.
Meeting Channel is part of Ways you talk to them.
Chat in the app with live reasoning, speak to your AI voice agent over a call, message on Slack or WhatsApp, send an email, or embed a widget on your website. Every channel connects to the same AI agent with the same memory and capabilities. Switch channels mid-conversation and your employee picks up right where you left off.