To be honest, AI Agents can already do quite a lot now: check-in, cross-chain, batch claim. Once you set the rules, you basically don't need to manage them. But when it comes to actually spending money, I still take a look.

It's not that I don't trust the code; it's that I don't trust the prompt I wrote myself. Last time, a bot set the gas limit too low, got stuck for three hours, and almost missed the window. In short, on-chain is irreversible. No matter how smart the Agent is, the final safety net still has to be a human.

Recently, ETF fund flows and the US stock market have been volatile, causing the overall market risk appetite to swing. At times like this, I dare not fully rely on automation. At least for large operations, I manually confirm first, and only delegate small tasks to scripts.

Anyway, my principle is: save time where you can, but bear the responsibility you cannot delegate.