That bot we have to spin up takes like two minutes to spin up in the cloud to be able to join your meeting. If you ask for the bot to join your meeting, you're not gonna wait two minutes. It needs to show up within ten seconds, which means I need a predictive algorithm.

to be on the bleeding edge means you're bleeding. You're cutting your hands on all these sharp edges, right, that are happening with Zoom, like, and that was our sharp edge like making this work seamlessly inside of these video conferencing platforms that had no real good APIs for doing this. So orchestrating all that in real time, you know, a lot of it was also like the capacity constraints, right? So that bot we have to spin up it takes like two minutes to spin up that bot, right, in the cloud to be able to join your meeting. If you ask for the bot to join your meeting, you're not gonna wait two minutes to for it to show up. It needs to show up within ten seconds, which means I need a predictive algorithm on how many bots I need to have on hand at every half hour increment. Right? Which means I need to look at people's calendars and make some estimations and da da da da. So there's a lot of stuff around that where it's like, yeah, if you wanna do a naive version that, like, shows up two minutes away, it takes thirty minutes to get you the transcription.