Actually, everyone understands that no matter how "automatic" an AI Agent is, there still needs to be someone overseeing the on-chain step; otherwise, if something goes wrong, you won't even know who to blame.

For someone like me who keeps an eye on Gas fees, I'm most afraid that it might get excited and send multiple transactions at once, causing congestion to spike, and the fees will be as high as peak taxi fares in real life— the more urgent, the more expensive...
So right now, I mostly let it do "watching": scan on-chain quotes, estimate slippage, and calculate whether a transaction is worth it;
When it comes to signing, authorizing limits (especially unlimited authorizations), cross-chain transfers, or verifying if the contract address is correct, I still do it manually, even if it’s slower.
Recently, some people are desperately hyping up the narrative while also starting to scrutinize security details.
I, on the other hand, lean towards the latter: automation is fine, but don’t automatically hand over your wallet.
That’s all for now.