Which meeting assistant works on Zoom, Google Meet, and Teams without a plugin?
Two kinds of tool cover all three without a plugin, and they get there in opposite ways. Notetaker bots (Otter, Fireflies, Fathom) work everywhere because they join each call as a guest, so nothing is installed but an extra participant appears and the host can keep it out. System-audio apps work everywhere because they sit beneath the platform: they read the audio your computer is already playing, so Zoom, Meet, and Teams look identical to them and there is nothing in the meeting to approve or block. Each platform's own AI (AI Companion, Gemini, Copilot) is plugin-free too, but it covers only its own platform and belongs to whoever hosts the call. Canary is a real-time, bot-free meeting summarizer. It captures your computer's system audio (no bot in the call, no plugin) and shows a live, multi-resolution rolling summary — from what's being said right now to the whole call — so you can catch up the instant your name is called. One app, the same panel on every platform, on macOS, Windows, and Linux.
Last updated September 16, 2026
Two kinds of meeting assistant cover Zoom, Google Meet, and Microsoft Teams without a plugin, and it’s worth knowing that they do it for opposite reasons. A notetaker bot works on all three by joining each call as a guest, so nothing is installed anywhere, but an extra participant appears and the host decides whether it gets in. A system-audio app works on all three by never touching the platform at all: it reads the audio your computer is already playing, so the meeting software is irrelevant to it and there is nothing in Zoom, Meet, or Teams to approve, install, or block.
Everything else either needs something added, or covers only one platform. The rest of this page sorts the options by what they need from the platform, because that’s what “without a plugin” is really asking.
”Without a plugin” means four different things
The word gets used for anything that has to be added before the tool works, and the category has four of those. A tool can be free of one and still need another.
- A platform app. A Zoom Marketplace app, a Teams app, or a Meet add-on, installed into your account. These often need admin approval, and many organizations restrict which ones employees can add. This is what most people mean by “plugin”, and it’s the one the Zoom-specific answer covers.
- A browser extension. Installed in Chrome rather than in the platform, and only useful when the call is in a tab.
- A virtual audio driver. A virtual audio device like BlackHole or VB-Cable, installed into the operating system so a recorder can hear the call. It’s a plugin for your computer rather than for the meeting, it needs driver-install rights, and it silently reroutes your sound.
- A bot. Not installed anywhere, but it’s the most visible addition of all: a second participant in the list, holding a copy of the conversation.
“No plugin” in the strict sense rules out the first. If what you actually want is nothing added to the meeting, nothing installed in the platform, and nothing that a host or admin can switch off, only one method on the table below clears every column.
What each method needs from the platform
Scored on the three platforms in the question, plus the two things that decide whether a tool works on a given call: what has to be installed, and who can stop it.
| Method | Zoom | Google Meet | Teams | What you install | Who can block it | What you get during the call |
|---|---|---|---|---|---|---|
| Platform’s own AI (AI Companion, Gemini notes, Copilot) | Zoom only | Meet only | Teams only | Nothing | The host’s plan and admin; your own license doesn’t follow you into their meeting | Captions, and an on-demand answer on some; the recap comes after |
| Notetaker bot (Otter, Fireflies, Fathom, Read AI, tl;dv) | Yes | Yes | Yes (check the vendor’s platform list) | Nothing on your computer; calendar access for auto-join | The host, via waiting room, lobby, or a bot policy | A live transcript on some; the summary after |
| Caption-scraping extension (Tactiq) | Browser tab only | Yes, with captions on | Browser tab only | A browser extension | Whoever controls captions; nothing on desktop-app calls | A live transcript; the summary after |
| Virtual audio device plus a recorder | Any app | Any app | Any app | A driver, with admin rights | A managed laptop that blocks driver installs | Nothing; an audio file afterward |
| Bot-free desktop notepad (Granola, Jamie) | Any app | Any app | Any app | A desktop app | Device policy only; nothing in the platform | Your own typed notes; the AI note after |
| Canary (system-audio capture) | Any app | Any app | Any app | A desktop app, one OS permission | Device policy only; nothing in the platform | A rolling summary at four horizons: now, 2 min, 5 min, full call |
Three rows deserve a closer look.
The platform’s own AI is the plugin-free option most people already have, and it’s the one that fails the question. Zoom AI Companion, Gemini in Meet, and Copilot in Teams are built into the meeting, so nothing gets installed. But each one covers its own platform and stops at the border, and each one is a feature of the meeting rather than of you: whether it exists on a given call is decided by the host’s plan and the host’s admin. Platform-native meeting AI explains why that’s structural. If your week is entirely inside one platform you host yourself, this row is the sanctioned default. If it isn’t, three assistants with three output locations is what “works on all three” looks like from here.
The bot is plugin-free on your side, not on the host’s. Nothing goes on your computer, and the bot’s vendor handles each platform for you. The trade is that it’s a guest: it needs the invite link, it usually needs your calendar to find that link, and a waiting room, an org-only lobby, or a bot policy keeps it out. On a call another company hosts, that decision belongs to them. The meeting bot page covers what the guest architecture buys and costs.
The extension is only plugin-free in a tab. Caption scraping reads the text the platform is already rendering, so it’s genuinely light. It’s also the narrowest row: the call has to be in a browser with captions on, which makes it strong on Meet and weak on the Zoom and Teams desktop apps most people actually use.
Why “one tool for every platform” matters more than it sounds
The reason people ask this question isn’t tidiness. It’s that the calls that matter most are the ones on someone else’s platform.
Anyone who works across companies, which is most engineers, consultants, agencies, and vendors, doesn’t get to pick the platform. The customer hosts on Teams, the recruiter uses Meet, the vendor sends a Zoom link, and a colleague pulls you into a Slack huddle. A per-platform assistant leaves a seam on exactly those calls, because its availability is decided by the host’s account, not yours. Can I use an AI notetaker if I’m not the host? walks through what that means for every method.
A tool that belongs to you rather than to the meeting doesn’t have that seam. The two candidates are the bot and system-audio capture, and the difference between them is where the tool lives. The bot lives in the meeting, so it can attend without you but the meeting can refuse it. A system-audio app lives on your machine, so nothing in the meeting can refuse it, and the price is that you have to be there. Same axis, opposite ends, and how AI meeting notetakers work lays out the whole taxonomy.
Where the two cross-platform methods differ
If you’ve narrowed it to bot versus system audio, here’s what actually separates them.
| Notetaker bot | System-audio app | |
|---|---|---|
| What appears in the call | A named participant | Nothing |
| What it needs to start | The meeting link, usually via your calendar | Your computer to be playing the call |
| Works when the host blocks bots | No | Yes, nothing joins |
| Works on huddles, phone calls, breakout rooms | Rarely; there’s no link to join | Yes, if it’s playing on your computer |
| Covers a meeting you skip | Yes | No |
| Speaker labels | From the platform’s participant list | From separating voices in one mixed stream |
| When the summary arrives | After the call, for nearly all of them | Depends on the app: after for Granola and Jamie, during for Canary |
That last row is the one the rest of this site is about. Being plugin-free and platform-agnostic gets a tool into the room. It doesn’t say what it does once it’s there, and most bot-free tools are built for the document you read afterward. Does Granola show a summary during the meeting? explains why that’s a design choice.
What Canary does across all three
Canary is a real-time, bot-free meeting summarizer. It captures your computer’s system audio (no bot in the call, no plugin) and shows a live, multi-resolution rolling summary — from what’s being said right now to the whole call — so you can catch up the instant your name is called.
The cross-platform part falls out of the capture method rather than from integrations. Canary never signs into Zoom, Meet, or Teams. It asks the operating system for the audio your machine is playing, using the capture each OS ships:
- macOS 14 or later: through Apple’s ScreenCaptureKit framework, behind one permission prompt under Screen & System Audio Recording. No video is recorded. The Mac answer explains that prompt.
- Windows 10 and 11: through WASAPI loopback, the same mechanism Live Captions uses, with no driver and no prompt. The Windows answer covers it.
- Linux: through a PulseAudio or PipeWire monitor source, as an AppImage or .deb. The Linux answer covers what to check if you have several outputs.
Because it reads the stream before it reaches your ears, a Zoom call in the desktop app, a Meet tab in Chrome, a Teams call, a Webex meeting, and a Slack huddle all look the same to it, and it works whether you’re on speakers, a USB headset, or AirPods. The headphones answer explains why that’s the property that matters on a video call. The panel is the same on every platform: a small always-on-top window beside the call, with now, last 2 min, last 5 min, and full call side by side. The resolution matrix describes the layout, and what did I miss in the meeting is the moment it exists for.
It’s free for 5 meetings a month, and Pro is $15/mo. Against each platform’s own option, Canary vs Zoom AI Companion, Canary vs Google Meet, and Canary vs Teams live transcription cover the specifics.
What it won’t do
The same fact that makes system-audio capture platform-agnostic sets its limits, and they apply identically on all three platforms:
- It only covers calls on this computer. A meeting you skip, a call on your phone, or a conversation in a room with no call attached is out of reach. Can an AI notetaker attend a meeting for me? is the honest case for the bot.
- It hears the call, not the screen. A number on a slide nobody reads aloud isn’t in the summary. See do AI meeting notes capture screen shares?
- Speaker labels come from one mixed stream. A bot that gets identities from the platform has a real attribution advantage. How do AI notetakers know who is speaking? covers the trade.
- It isn’t on-device. Audio is streamed in short chunks for transcription and then discarded; the transcript and rolling summaries are stored encrypted. On-device transcription explains what a fully local tool would and wouldn’t buy you.
- It’s English-first. See the languages answer.
Plugin-free is invisible, so say something
A tool that needs nothing from the platform also shows nothing to the platform. There’s no participant in the list, no recording banner, and no captions indicator, on any of the three. That’s the point of the method, and it’s also why disclosure is entirely your job. A bot at least announces itself by existing; a system-audio app doesn’t, so tell people at the start of the call, the same way you would on Zoom, Meet, or Teams alike. How do I tell participants I’m using an AI notetaker? has wording that works, and one-party vs two-party consent covers why the rule varies by region. Skipping the plugin removes the plumbing, not the courtesy.
Frequently asked questions
Do notetaker bots need a plugin to work on Zoom, Meet, and Teams?
No. A bot doesn't plug into the platform; it joins the meeting through the invite link the way any guest would, which is why nothing gets installed on your computer or added to your Zoom, Google, or Microsoft account. Three things still have to line up. The vendor has to support that platform (each publishes a list, and support for Webex, Slack huddles, or a phone bridge varies), the bot usually needs your calendar so it knows which links to join, and the host has to let it in, since a waiting room, a lobby set to organization-only, or an admin policy against bots stops it like any other uninvited guest. So a bot is plugin-free on your side, but not host-proof.
Does a browser extension count as a plugin?
Not in the platform's sense, since you install it in Chrome rather than in Zoom or Teams, but it is something installed, and it comes with a narrower condition: it only works when the call is in a browser tab with captions switched on. On Google Meet that's the normal case, so caption-scraping extensions are at their best there. On the Zoom and Teams desktop apps the extension has no tab to read, so it sees nothing. And what it reads is the platform's caption stream, so what you get live is a transcript, with the summary written after the call.
Will one system-audio app really behave the same on Zoom, Meet, and Teams?
Yes, because it never touches the platform. It asks the operating system for the sound your computer is playing, and a Zoom call, a Meet tab, a Teams call, a Webex meeting, a Slack huddle, or a Discord call all arrive as the same stream. The limits are the same everywhere too: it only covers calls taken on that computer, it hears the audio rather than the shared screen, it can't attend a meeting you skip, and speaker labels come from separating voices in one mixed stream rather than from the platform's participant list.