Meet the Meeting was built to solve a familiar problem in online presenting: when you speak into a grid of silent video tiles, it is hard to tell whether your audience is with you or drifting away. The extension taps into the video streams already visible inside a Google Meet call and runs a lightweight computer-vision model on them locally, summarising the mood of the room as a table of expression counts and percentages rather than as a wall of raw feeds.
Because the heavy lifting happens through Web AI in the browser, the developer states that no video or expression data is sent to external servers. That design choice keeps latency low and makes the tool easier to justify in workplaces where meeting content is sensitive. The project is released under the Apache-2.0 licence and its code, including the bundled face-detection model, is published openly, so technically minded users can inspect exactly what runs and even fork it.
The tool is deliberately small in scope. It is a single-purpose Chrome extension rather than a full meeting-analytics suite, it is free to install, and its audience is presenters, teachers, and facilitators who want a lightweight, private read on how a Google Meet session is landing. Users should treat the output as a rough signal, not as a precise emotional measurement, and should always consider consent and privacy norms before analysing other people's faces.
