The basics of building the Meeting Bots were quite simple: we use Puppeteer to log into the meeting and scrape its content for our users.
What was challenging was scaling these bots using Kubernetes and ensuring they are 100% reliable. Any small change in the video meeting provider interface could lead to unreliable data being scraped.
To help any developer get started quickly, we have also open-sourced a recording interface that lets you view recordings, chat with the transcript using an LLM, and more.
https://meetingbaas.com/transcript-seeker/
The latter is also compatible with transcription APIs, and I plan on making it compatible with Whisper and then package it inside a Tauri App - to have (at last!) an open-source transcription app that allows text-editing, SRT download, and more.
Happy to answer any questions you might have on how we built this!
No comments yet.
