Meeting bot

A meeting bot (also called a notetaker bot) is an automated participant that an AI meeting tool sends into a video call to record it. The bot joins through the meeting link the same way a human guest would, takes a seat in the attendee list — usually with a label like “Otter.ai Notetaker”, “Fireflies.ai Notetaker”, or “Read.ai” — and streams the call to the vendor’s cloud, which turns it into a transcript, summary, and action items after the meeting ends.

How a meeting bot works

Most bot-based notetakers follow the same loop:

1. Calendar connection. The user connects the notetaker to their calendar. The tool watches for events that contain a Zoom, Google Meet, or Teams link.
2. Auto-join. At meeting time, the vendor’s servers spin up an automated client that opens that link and requests to join — exactly like a guest clicking the invite. Nothing runs on the user’s computer; the bot lives in the vendor’s cloud.
3. Capture. Once admitted, the bot sits in the call and streams the audio (and often video and screen shares) back to the vendor for streaming transcription and summarization.
4. Delivery. After the call, the owner gets the transcript and summary by email or in the vendor’s app, sometimes shared automatically with other attendees.

Because the bot is a participant, the platform treats it like one: it can be held in a waiting room or lobby, admitted, muted, and removed by the host.

Why notetakers use bots

The bot architecture is popular for practical reasons:

• It works from anywhere. The user can join from a phone, a tablet, or a conference room — the bot captures the call from the cloud regardless of what device they’re on.
• It can attend without you. If the owner skips the meeting, the bot still joins and takes notes — the “send my notetaker” workflow.
• Cleaner speaker attribution. The platform tells participants apart for it — name tiles and active-speaker signals give the bot reliable speaker labels, a shortcut around harder acoustic speaker diarization.
• It sees the room. A bot can capture video and shared screens, not just audio.

The tradeoffs

The same architecture is why bots have become the most visible irritation in meeting AI:

• It’s an extra participant. Everyone sees it, every meeting, which changes the feel of client calls and prompts the perennial “why did a bot join my meeting?”
• It can be blocked. Waiting rooms, lobbies, and org-level admission policies stop bots the way they stop any uninvited guest — many companies block them deliberately, and hosts can simply not let them in.
• Bots multiply. If three attendees use three different notetakers, three bots ask to join the same call.
• The record lives elsewhere. The recording and notes sit in the bot owner’s vendor account, outside the control of everyone else on the call.

Bots are one capture method, not the category

A meeting bot is one of four ways AI notetakers capture a call — alongside platform-native AI (Zoom AI Companion, Teams), browser extensions that scrape captions, and system audio capture, which reads the sound your own computer is already playing. The full taxonomy is in how AI meeting notetakers work; the bot-free branch is covered in the complete guide to bot-free meeting notes.

One thing bots get right is visibility: a labeled participant is a form of disclosure. Going bot-free removes the extra guest, not the courtesy — you should still tell participants you’re taking AI notes. Canary takes the bot-free path: it captures your computer’s system audio (no bot in the call, no plugin) and shows a live, multi-resolution rolling summary during the meeting — the opposite of the bot’s record-now, summarize-later loop. See how that plays out against a bot-based tool in Canary vs Fireflies.