Platform-native meeting AI

Platform-native meeting AI (also called built-in or first-party meeting AI) is a meeting-capture method in which the video platform’s own assistant — Zoom AI Companion, Microsoft Teams intelligent recap, Google Meet’s Gemini notes, Webex’s AI Assistant — transcribes and summarizes a call from inside the meeting itself, with no third-party tool involved. Nothing joins the call, nothing installs on your computer, and the notes are produced by the same company that runs the meeting.

It is one of the four ways an AI meeting notetaker can get hold of a conversation, and the only one that works from inside the meeting rather than looking at it from outside.

How it works

There is no capture step to bolt on, because the platform already has the audio:

1. Someone switches it on. Depending on the platform, that’s the host starting the assistant, or a participant with a license invoking it — on top of whatever an administrator has already allowed for the organization.
2. The platform transcribes its own media. It has each participant’s stream separately and knows who is on the roster, so speaker labels come from identity rather than from acoustic speaker diarization.
3. A model summarizes the transcript. Usually once the meeting ends, producing a recap with decisions and action items.
4. The output lands in the platform’s own storage. A summary in the meeting’s chat or recap tab, a notes document in Drive attached to the calendar event, a recording in the platform’s cloud — inside the account your organization already administers.

Availability, packaging, and feature names shift often and vary by plan, edition, region, and admin policy, so treat any specific arrangement as something to check on your own account.

What follows from being inside the pipeline

It has the best raw material of any capture method. Separate per-participant streams, names from the roster, and access to video and shared screens. A tool listening from outside is working with less.

It can cover a meeting you never opened. Like a meeting bot, and unlike anything that runs on your machine, the platform can produce a recap of a call you skipped entirely.

It is the sanctioned option. It’s already covered by the contract your company signed with the platform vendor, so there’s no new vendor, no procurement, and no software to install — which matters on a managed laptop where bots or app installs are blocked. Retention and eDiscovery are handled by whatever policy already governs the platform.

Disclosure is largely automatic. Platforms announce themselves — a banner, a notice on join, a note in the chat. Participants generally find out that AI is running without anyone having to say so.

It’s usually not an extra subscription. On many paid plans the assistant is included or available as an add-on to something you already buy, which is a real advantage over paying for a separate tool.

It’s a feature of the meeting, not of you

This is the defining property, and everything awkward about the method comes from it. Every other capture method belongs to the person and travels with them. Platform-native AI belongs to the meeting — so whether you get notes is decided by whose meeting it is.

Three separate parties have to line up, and none of them is necessarily you:

• Your plan or license. The edition your organization bought, and whether an AI license is assigned to you.
• An administrator. Tenant- or account-level settings that enable the assistant, and sometimes prerequisites like transcription being permitted at all.
• The host. On most platforms someone has to actually start it, and that someone is whoever created the meeting.

On an internal call, those three are usually all on your side. On a call another company hosts, all three sit with people you don’t work with, your own subscription doesn’t follow you in, and the answer is often simply no — which is why using an AI notetaker when you’re not the host is a genuinely different question. The calls where you most need notes — customer calls, vendor calls, interviews, client work — are disproportionately the ones you didn’t create.

It doesn’t travel between platforms

Each platform’s assistant covers that platform and stops at its border. Zoom’s is blind to your Teams calls, Copilot doesn’t see Zoom, and Slack’s huddle notes cover huddles only. If your week spans several platforms — the normal case for anyone working across companies — native AI gives you a patchwork of coverage with different feature sets, different output locations, and seams falling exactly on the cross-company calls.

Most of it is after the meeting

Native meeting AI is overwhelmingly a post-meeting product: the summary is written once the call ends. What exists during the call is usually a live transcript — captions — which is not the same thing as a live summary.

The in-meeting versions that do exist are on-demand rather than ambient. Zoom and Teams let a licensed user type a question into a chat panel to ask what’s been covered. Webex’s Catch Me Up is the closest any platform has come to a genuine mid-meeting catch-up, and Google’s note-taking can summarize the discussion so far for someone who joined late. All of them share a shape: you have to notice you’re lost, stop, and ask — one step too slow in the moment you were already behind, and a single answer rather than a view you can glance at again thirty seconds later. The platform-by-platform detail is in the live-summary answers for Zoom, Teams, and Google Meet.

Where it sits among the four capture methods

• Meeting bot — an automated participant joins through the link. Device-independent, but visible in the attendee list and blockable.
• Platform-native AI — this page. Best fidelity and already sanctioned, but gated by plan, admin, and host, and confined to one platform.
• Caption scraping — a browser extension reads the captions the platform is already drawing. Nothing joins, but it’s bound to the tab and to the platform’s caption stream.
• System audio capture — a desktop app reads the sound the computer is already playing, so it’s indifferent to which app the call is in and needs no captions, no bot, and no plugin.

The last three are all bot-free; they differ in what they’re bound to. Native AI is bound to the platform and its permissions, an extension to the browser tab, system audio to the machine. The full pipeline behind all four is in how AI meeting notetakers work.

The transparency point

Platform-native AI is the most transparent method in the category, and that’s worth saying plainly: the platform tells everyone the assistant is running, so nobody has to remember to. Any tool that runs outside the platform — bot, extension, or desktop app — inherits that job. Removing the extra participant removes a guest, not the courtesy, so tell participants you’re taking AI notes. Recording and consent rules vary by region regardless of which tool produced the record.

When a standalone tool makes more sense

If you host nearly all your meetings on one platform, your organization has the assistant enabled, and you read notes afterward, the built-in option is the right default — it’s free-with-your-plan, approved, and better-informed than anything outside the call.

Canary is for the other case. It 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. Because it reads the machine rather than the meeting, there is no plan, no admin, and no host in the loop, and it works the same on every platform, including the ad-hoc calls that never had a calendar event. And because the summary is built while the meeting is still happening, it helps during the call rather than after it.

For the head-to-head versions, see Canary vs Zoom AI Companion, Canary vs Microsoft Copilot, Canary vs Webex, and Canary vs Google Meet — or the complete guide to bot-free meeting notes for the whole landscape.