Recently, everyone has been hyping up AI Agents for fully automated on-chain operations, and I’ve actually become a bit hesitant… To be honest, automation itself isn’t hard; what’s difficult is who takes the blame if something goes wrong. For example, authorization issues, whether the contract address has been changed, sudden slippage/router issues, or temporary glitches in cross-chain bridges—these are things I still need to confirm myself, at most having it list out candidate options first. And then there are position and stop-loss conditions; an Agent can easily “execute according to rules” until the account is wiped out, so a human needs to have the final backup.

Additionally, recently some regions have been tightening and loosening taxes and compliance measures, making deposit and withdrawal expectations quite unpredictable. I’m now even less willing to hand over all my funds to a robot. Anyway, lowering my expectations has made me feel more relaxed: the Agent is just a tool, and humans are the brakes. Let’s keep it like that for now.