Had this wild conversation with Evan Cheng from $SUI that honestly shifted how I think about all this.
We started joking about home robots — like imagine you come home one day and instead of cleaning your house, it's literally cleaning OUT your house. Walking away with your stuff. We laughed. Then he said "if you really think about it, it's entirely possible." That stopped me.
Evan's been around — built dev tools at Apple and Meta before starting Mysten Labs. We recorded this in Singapore and what started as a chat about stablecoins turned into something way bigger.
His main point keeps replaying in my head:
"The web is built for humans. You cannot retrofit it to handle agents."
Everything online assumes there's a person on the other end. Identity. Trust. The fact you can sue someone. Agents break all of it. I told him I get phishing emails daily that even make ME pause (and yeah, I confessed to 50,000 unread emails). Now imagine your agent reading those with full access to your accounts. His response: "Email is the biggest freaking attack vector... such a legacy protocol."
And then the most honest thing he said:
"Disaster will happen. That's part of the progress."
Look around — PHP sites aren't safe anymore, people are hacking fridges and freezers. Our whole infrastructure is outdated. This isn't some distant future problem.
We covered stablecoins, AI agents going rogue, who pays when your bot gets hacked, insurance for agents (?), why the internet will break under AI load, and how $SUI is trying to build for this new reality.
Honestly one of those conversations where you walk in with simple questions and walk out rethinking everything.
We started joking about home robots — like imagine you come home one day and instead of cleaning your house, it's literally cleaning OUT your house. Walking away with your stuff. We laughed. Then he said "if you really think about it, it's entirely possible." That stopped me.
Evan's been around — built dev tools at Apple and Meta before starting Mysten Labs. We recorded this in Singapore and what started as a chat about stablecoins turned into something way bigger.
His main point keeps replaying in my head:
"The web is built for humans. You cannot retrofit it to handle agents."
Everything online assumes there's a person on the other end. Identity. Trust. The fact you can sue someone. Agents break all of it. I told him I get phishing emails daily that even make ME pause (and yeah, I confessed to 50,000 unread emails). Now imagine your agent reading those with full access to your accounts. His response: "Email is the biggest freaking attack vector... such a legacy protocol."
And then the most honest thing he said:
"Disaster will happen. That's part of the progress."
Look around — PHP sites aren't safe anymore, people are hacking fridges and freezers. Our whole infrastructure is outdated. This isn't some distant future problem.
We covered stablecoins, AI agents going rogue, who pays when your bot gets hacked, insurance for agents (?), why the internet will break under AI load, and how $SUI is trying to build for this new reality.
Honestly one of those conversations where you walk in with simple questions and walk out rethinking everything.