Recently, everyone has been talking about AI Agents going on-chain to do work automatically, basically saving a little bit of effort, but when it comes to full custody, I'm still hesitant. Checking contract permissions, seeing if they can transfer your money arbitrarily—no matter how smart the agent is, someone still needs to verify and confirm; and then there's the signing step, once signed there's no regret medicine, you have to keep an eye on it yourself. As for the testnet points expectations and guesses about whether the mainnet will issue tokens... the agent might be good at completing tasks, but if the rules change at the end, all that effort is wasted, and it really messes with your mindset. Recently, I’ve been practicing trading psychology: don’t treat "automation" as certainty, what you can do is think through the worst-case scenario for each step, and avoid getting overly excited with wishful thinking.