Andy Halliday

mid-aughts to do that kind of QA.

And so ultimately, you had to say, OK, well, it seems like it's working well enough.

Let's give it out to the early adopters and get feedback from them.

Well, now it's a little different.

Like when they release something to their premium tier,

They definitely want to do risk reduction by ensuring that there's not going to be some crashing problem that they didn't identify in advance.

But automated AI QA testing is a thing.

You can imagine that anything that you're going to put in front of the public, you can have AIs bang through it a million times and see whether the code fractures anywhere along that path.

So I think you're not really exposing yourself to a jagged experience as a max user using deep research.

I'm sure what you would get right now as one of those early adopting users of advanced deep research is something that's better than the existing deep research.

Speaking of nano banana, researchers from Peking University and Google Cloud AI just released paper banana.

Yes.

System of five agents that work together to generate scientific research publication ready diagrams and charts for academic papers.

So, you know, one of the things that's kind of a little interesting.

incomplete about doing infographics or charts and so on with a nano banana is you'll see some errors in there, but overall it does provide some visual enforcement or reinforcement of the ideas that are in whatever information it is you're wanting to present.

But this system is, is creating and they show examples on the announcements really complex, but you know, high fidelity diagrammatics for,

you know, academic research.

So paper banana is, is happening and probably ultimately open AI with their prism science paper oriented, you know, research oriented platform.

We'll have to add some kind of image generation that's, that's that proficient as well.

Okay.