Medical AI: Choosing Post-Call Over Real-Time (Full Transcript)

[00:00:00] Speaker 1: So, I think both of you mentioned you're kind of doing like post-call analysis right now. I think a lot of people in the room might be building with like real-time. Maybe how do you make those decisions between like the real-time we were watching on the screen versus post-call? Are they used for different use cases within, you know, your application, your industry? Maybe give some insight to folks here like maybe why you made those as design choices and how you keep continuing to like look at them over time.

[00:00:22] Speaker 2: Precision here is of key importance in medical context and for us that's why it's not kind of live but like it's kind of processed afterwards as well and we kind of double check it and then we also make a clinician approve it. It's exactly because like you can't make mistakes. Like I mean from things like your membership ID to the patient's surname, these are like key informations. If you get them wrong then kind of the whole system collapses. So, for us that's why we like still even though there are more and there are more reliable models than there were even half a year ago, it's still I would say too dangerous for us to risk it.