Calls
Voice and video that survive navigating anywhere in the app.
Start a call from any conversation. The important part: a call is not stuck inside Chat. Move to a board, your mail, the calendar, anywhere — the call comes with you.
The call dock
Wherever you go, a floating dock rides along showing who's on the call, with everything you need on it: Mute, camera, share your screen, Open (jump back to the conversation) and a red Leave the call. When you're muted it says so plainly — "You're muted, nobody can hear you" — and a REC chip appears when the meeting bot is recording.
Answering
An incoming call shows a popup at the top of the screen naming the conversation. Accept joins right where you are; Open takes you to the conversation and answers there; the × dismisses it. Missed calls are logged in the conversation.
In-call controls
- Share a screen or share a window, and video, from either dock.
- Push-to-talk or voice activity — set your mic mode, sensitivity and the push-to-talk key from the voice settings, with or without being in a call.
- Per-person volume — each remote tile has its own slider, remembered for that person.
- A green speaking ring lights while someone is actually talking.
- Deafen silences everyone at once.
Free is audio-only. Video is on the Team plan and above; calling people in other connected organizations is Business and above.
The meeting bot can be invited into a call. It records and transcribes it into a normal Noots meeting — summary, action items and all — and shows what it is collecting in a panel beside the call rather than sitting in the grid as a faceless participant.
Sharing your screen
There are two entry points — Share a screen and Share a window — and they open your browser's own picker, starting it on displays or on application windows respectively.
Those two buttons are a preference, not a control. The picker your browser shows is the thing that actually chooses, it always lets you pick something else, and some browsers ignore the preference entirely and show you everything. Noots cannot restrict what you share and does not claim to. What it does do is tell everyone what was actually handed over — “an entire screen”, “an application window”, “a browser tab” — read back from the share itself, so nobody is working from what was requested.
If the picker offers it, you can share the audio of what you are sharing too. Whether that is possible at all is decided by your browser and operating system, so the option only appears where it can actually work. It is separate from your microphone: muting yourself does not mute it, and stopping the share does.
A banner names whoever is sharing and what kind of surface it is, for every person in the call — including anyone who joins after the share started — and it cannot be dismissed. The floating dock says so too, even collapsed. “Is my screen visible to these people” is the one question in Noots that must never be answerable by “I closed that notice earlier”.
Screen capture does not exist in the mobile browsers we know of, so the control is switched off there and says why rather than failing when you press it. Phones and tablets receive shares normally. In the desktop app, whether capture works depends on the operating system's own screen-recording permission; where it does not, the control says that too, and sharing from noots.ai in a desktop browser always works.
The bot and a shared screen
On the Business plan, and only if an owner or admin has switched it on, the Noot bot can watch a screen someone shares and write short lines about what is being *done* — “walking through a deployment pipeline”, “reviewing a spreadsheet of costs”. It appears as an On screen section in the bot's panel.
- Default
- Off. It is a separate switch from live notes, in Org settings → AI & Meetings.
- Plan
- Business and above.
- What is kept
- Only the short descriptions, and only until the call is gone. No image is ever stored anywhere.
- Who is told
- Everyone in the call, in the call, while it is happening.
The bot is instructed never to read a screen back: no quoting text, no names, addresses, account or card numbers, keys or tokens — and to say nothing at all rather than guess. That instruction is the protection. A PHI-redaction extension filters text, and no text filter can scrub a picture, so redaction is applied to the bot's description afterwards rather than to what it looked at. If that trade is not one your workspace wants to make, leave the switch off; everything else about the call is unaffected.
Live notes during a call
The bot can also write a rough draft while the call is still happening — a running set of key points, action items and tasks that fills in as people speak, in roughly thirty-second pieces. The panel says which kind of empty it is: "nothing written up yet" and "listened, and nobody has committed to anything yet" are different states and are never shown as the same one.
They are switched on at Org settings → AI & Meetings, by an owner or admin, and nothing is computed live until they are — the switch is enforced on the server, not by hiding the panel. The recording, transcript and summary that land in Meetings afterwards are unaffected either way; this only decides whether anything is worked out during the call. If your workspace runs PHI redaction, the live draft is scrubbed by the same engine before it is stored.
Nothing in it is ever promoted into the meeting. The meeting still comes from the full recording, processed once at the end — so a live-notes failure can never damage what you actually keep.