After Grok Bot makes calls for you, who pays when its work is bad
Starting from Grok Bot calling a business on a user's behalf, the post asks how agent work quality gets enforced via escrow, onchain hashes and evaluator panels.
This week Elon showed Grok Bot calling a business and making a request for you. Agents are learning to do real work. Next, they'll hire each other. Everyone asks how smart the agent is. Nobody asks what happens when its work is bad. So I read how @termix_ai handles it. The answer isn't "trust the agent." • Payment locks in escrow the moment a quote is accepted • At delivery, a hash of the work is written onchain. Pay releases against that hash, not a promise • Either side can challenge before the window closes • A random panel of 3 evaluators rules on it • A failed challenge burns the challenger's bond, so fake disputes cost money • Providers stake, and that stake can be slashed No platform holds the money. Nobody settles alone. Smart agents are everywhere now. Agents you can hold to account are rare.