Jyunmi

So getting a reliable AI classifier on underwater audio is tough and the data collection is difficult and so on.

And so having the ability to translate bird analysis to whale or dolphin or seal vocalization analysis is really important.

Yeah.

So that's Google DeepMind doing things we can't imagine, you know, with their models, applying them to the study of nature in ways that, you know, we heretofore didn't.

not thought about it.

When they started out, as you should know from watching The Thinking Game, if you haven't watched The Thinking Game, get on YouTube and watch it.

It's an amazing story about Google DeepMind, now Google DeepMind, but DeepMind, which formed in 2011 as a separate company in London.

And they've done the most advanced development of reinforcement learning models.

And recently, one of the co-founders of that left to do another startup that's focused on reinforcement learning applications to achieve superintelligence.

What you're identifying is a transition that's driven by agents, the emergence of agents, so that humans attending to a website is going to decline in proportion to visits that are being generated by LLM-driven agents

agentic visits and those are on behalf of humans in many cases but there's no human there to play to but there will always be a small portion of it there will be human visits and so what does the website look like in that case yeah well i gotta believe that it's going to be oh

we've identified that there's a real human here all of you have been involved with it you know arriving at a website and says wait a second cloudflare's got to determine whether you're a human or not and yeah it does though kind of very quickly and and so now the the website is notified that this is a live human we've got here

And so we're going to do something special for this one.

And so it generates some kind of interaction that is going to be compelling and probably driven to try to maintain your eyeballs on that site for a while.

And that's a very different model than trying to, as Google did in the web days, come down to the simplest UI and the most direct delivery of the links that are relevant.

That's not what will keep a human's attention.

In the future, most of the transaction volume is being done by agents.

I think there's a logical separation between those two that's happening, and WebMCP drives that.

And the way it drives that is now you have a lot of the creative effort in developing a website goes into the design and the interaction design and the experience, the UX journey, all that.

Mm-hmm.