Last night, while munching on skewers, I watched a few AI agents running on the chain, and it feels like they are now more like "executors." When it comes to making decisions, humans still need to step in as the final authority. For example, giving too much permission makes it easy to fall into phishing contracts; giving too little causes delays in signing, changing parameters, or stop-loss actions. The most annoying thing is when they "think they are right," but you have no idea why they are doing it, and if something goes wrong, you can only blame yourself for granting the authorization in the first place.

By the way, I also thought about the recent NFT royalty debates... Basically, it's a game of automation rules versus human nature: you want the contract to automatically protect creators' income, but secondary liquidity doesn't buy into that. In the end, it's still the platform, project teams, and buyers and sellers fighting over it, with the contract just taking the blame.

My partner also complained about me: you watch the chain as closely as you watch food delivery progress, but when the agent places an order with one click, you don't dare to let go... Well, indeed, I prefer to take it slow with money—let's keep it that way for now.