Lei Yang

It's fast because you are actually talking to data center that is maybe five or 10 milliseconds away from you.

Interesting question.

So yeah, I personally do use AI to code.

And so I think one, and I also try to automate the company using AI, but I failed because I have big trust issue.

So for example, I, yeah, really.

So I have big trouble trying to connect OpenClaw or any kind of cloud hosted AI to our company Slack.

just because I don't want some other company or I don't want information to be transmitted to yet another company.

I know Slack is cloud-based, so I'm kind of paranoid.

But anyway, so I think the same paranoia can be applied to blockchains because if you think about AIs, they are interesting.

They are great machines that can propose solutions that are 99% of the time perfect.

but 1% of the time catastrophic.

So how do you make that work for your chains users?

So my mental model has been to comparing AIs or AI agents with very talented intent solvers.

They work very similarly.

So we'll tell an intent solver in say an intent protocol, hey, please find me the best way to swap, say USDC into Ethereum, into Ether.

You care about a quote and also, of course, you care about security.

For example, the intent solver should not be able to take your money away, but you do not really care about how they arrive at the amazing quote they actually materialize, right?

I think same goes for AI.

When you tell AI to code, I think...

people start to care less and less about what actual code was written.