Mm-hmm.

So when a cycle, a rate environment is decre-declining in property or natural catastrophe insurance, there are other lines of businesses that are not correlated to that, and I think that's what we observe.

We look at our total portfolio, and we call it the target liability portfolio.

We have a five-year forward-looking view, and we assume devel-developments, rate developments, but also inflation, et cetera.

That all comes in, and then we see, how does my portfolio behave?And if there's one cycle that goes down, I probably will not grow in that area so much anymore.

So I limit capacity deployment to that area.

Mm.

So that's where, where you can then detract the benefits because the problem very often is that data and technology is outdated, it's fragmented, huge IT legacy and IT debt so that the individual AI use cases are still standalone use cases and cannot be integrated into your data and, uh, technology infrastructure.

Mm-hmm.

With AI, we add AI basically, and we augment the offering on this lifeguard side.

So obesity and all the, uh, uh, uh, specific aspects of it obviously are entered into our model and into our assumptions obviously.

And, um, we had this, this debate around COVID.

You know, what was the reason, the causality for these spikes, spikes that we could see after the core event of COVID? Yeah, so the late reporting of claims and we had in...

I think it was in 2023, in Q3 if I remember correctly, we had a spike coming, spike coming through from the US.

And we were scratching our heads and we say, "Why is this happening in the US, in particular in the age group, um, uh, of middle-aged people?" Because in all OECD countries, you didn't see that spike, and that causality needed to be understood.

Mm.