Someone asked me, "Is putting AI agents on the blockchain the end of needing humans?"… I’ll just laugh for a bit. If I hand over all wallet permissions to a robot, I bet the very next second it will learn what “authorization is forever” means. Basically, the most critical things humans still need to oversee are a few steps: clearly see who the signature/authorization is given to, whether the limit is unlimited, whether to use cross-chain bridges, and whether the contract is a newly deployed one. Robots can calculate paths to save gas, but they don’t take responsibility if you make a mistake and everything goes wrong.

Recently, I’ve also grown tired of the L2s arguing over TPS/fees/subsidies. Cheap is cheap, but as interactions increase, signature pop-ups become like whack-a-mole… Without human oversight, I don’t feel at ease. Anyway, I now only let agents do queries and simulations. If real money is involved, I still confirm myself—slow down if needed, life is more important. You say, “That defeats the purpose”… Let’s leave it at that for now.