Logan Kilpatrick

having to pull in and out or get distracted and go check your phone while the model is doing something or do some other random task and lose track of it.

So I think we're going to see that, which is really interesting.

And I think about this a lot for AI products.

But how do you keep people engaged, given that the models inherently have this latency, not problem, but characteristic about them?

So I'm excited.

I'm curious to see how that trend goes.

And I think the fun thing is like it can flash can be your daily driver model and it's free and it's like available everywhere and all that stuff, which is really cool.

So you don't have to be like paying $2,000 a month or $200 a month for one of these like pro subscription products.

You can just like use the default experience across many products in the industry and ecosystem and sort of have that frontier intelligence, which is really, really cool.

This is a good question.

I made this comment when we launched 3D Pro, which is, I think these models are now sort of turning the corner of how you have to think about working with AI systems.

And I think historically you would sort of, you know,

how can I make the most perfectly crafted thing, this whole notion of prompt engineering.

I need the model to do this very specific thing.

I have to artfully explain what I need, all this stuff.

If you mess up a couple of words here and there, maybe it won't do the thing that you need it to do.

You would also principally give it the smallest bite-sized chunk possible.

Do this one very specific thing.

Don't do anything else, please.

it goes on lists and things and very yeah and now and i i think that's changed like now i'm like um and i had this conversation with sundar and we were talking about like an example of a product that he he wanted he was like okay well i'll just give you the five things uh i really have these 30 things that i want but um you know i'll tell you five for the sake of this example and i was like no you got to tell me the 30 because like actually that's what that's how capable these models are