Lately, I've been hearing everyone talk about AI Agents automatically running on-chain interactions, and I'm quite tempted. After all, it can save a lot of effort on monitoring the market and tasks... But once I actually try it, I realize that some steps still need human oversight, or else the "snapshot anxiety" can easily turn into "zeroing out anxiety." Especially the authorization part—if the agent gets too excited and gives you unlimited approvals, you won't even know where you lost money if something weird happens with the contract later; and those signature pop-ups, many look exactly the
View Original