So what questions, um, help you uncover what the kind of real situation is?

Yeah.

Yeah, of course.

So I think there are, there are a few things that, that, that we would look at, right? So one is, um, one is a kind of, again, as I mentioned, discrete, uh, individual tasks or concepts or deployments versus a bold workflow information.

So I think it's, it's very important when you're, you know, if we are doing the due diligence or if the buyer is kind of doing the due diligence themselves, for them to be able to understand how does this business work end to end, um, across different functions within the business, and what is actually using AI at the moment and what is not.

And is there a real potential for the AI to literally change the entire workflow, transform the entire workflow end to end, um, for those, uh, essentially expectations of time freed or, you know, uh, increased margins you get.

So that's, that's the first one.

But how are we gonna train those humans?

Now, when you talk about the flashy presentations and then your own questions, so what questions help you uncover what the kind of real situation is?

Yeah, yeah, of course.

So I think there are a few things that I would look at, right? So one is...

One is a kind of, again, as I mentioned, discrete individual tasks or concepts or declaimants versus Apple workflow information.

So I think it's very important when you're, you know, if we are doing the diligence or if the buyer is kind of doing the diligence themselves for them to be able to understand how does this business work end to end across different functions within the business.

And what is actually using AI at the moment and what is not? And is there a real potential for the AI to literally change the entire workflow or transform the entire workflow end-to-end for those essentially expectations of time-free or increased margins? So that's the first one.

The second one, we would say, is data, right? There is a concept of, of course, available data, but then there's also a concept of usable data.

But how are we going to train those