And so we try to distill those use cases down into something that works within those constraints in the platform and then try to figure out Is this a primary use case? How much magic should happen? How much manual effort? How much awareness do you want to give to the user versus just kind of, again, doing all the things magically in the background? And so, again, I think that's another case where it could be something pretty internal in the guts, but there's still a back and forth with the analyst to understand how to kind of fit that use case.

There's also kind of the interesting intersection there of like engineers using the platform to do things and analysts using the platforms to do things where the analysts might want to be hand bespoke doing their analysis and curating their data sets.

Then you have engineering shaped users that are, I want to try to throw a million things into synapse every five minutes.

And they're wondering why things break down after doing that for nonstop for three weeks straight.

Um, and it might come down to like, I don't necessarily understand some technical aspect of the platform.

Um, and we might have to explain to them, Oh, your queries are actually grown or like N squared.

Cause you're, it's, they were fast when you started.

I think it's maintained that same level of quality and performance with extra bonus stuff added in, which I think is a big testament to our engineering team.