You know Grain and Firefly bots that attend the call to later generate notes and all that?!
I’ve been commissioned to build something similar but the difference is it should make notes in real-time and make some suggestions of follow-up questions and all that.
I’ll be using instant transcription to capture the audio and every time it hits 300+ characters, send that to OpenAI. So it can generate suggestions.
However what I’m worried about is how to handle incomplete sentences or thoughts?!
Is there any service that automatically detects where the speech concludes?!
Open to suggestions