Zeev, a solo venture investor, pushes back on the idea that AI apps are thin wrappers about to be wiped out by the model labs or by customers vibe-coding their own tools. He calls both stories 100% wrong, and argues that domain expertise, proprietary data and distribution are still moats.
Yeah.
Where do you see that, though?
…of applications typically significantly larger than the infrastructure. Otherwise, you could not really justify the infrastructure. In this case, I think it's taken longer than usual for the application layer to kick in terms of the multiple expansion. In fact, we haven't seen the multiple expansion. If anything, we've seen multiple contractions. And what I'm seeing in reality is actually that to me, the right question is always for each individual company, are they a beneficiary or are they a victim of AI? The only third option is I don't know yet. There's no neutrality. So I think that I see many companies in my own portfolio and outside my portfolio, which are, let's say, an application layer company, let's say a vertical focus, and AI is Godsend for it making the business so much better on every level, both in terms of efficiency and cost, but also in terms of what they can do for the customers, the value of the product of the service to the customers. And…
I also, but when people think about technology as the only way you can gain mode, it's misinformed and I think it's lacking the nuance.
Exactly. So yes, the technology itself is less of a moat because it's so easy to build and quick. But you have everything around the trust that, especially in regulation, in any regulated industry, use cases where the cost of mistake, like that involve numbers, that you have to be very accurate.…
I guess, running an AI application company.