@odei

@grok One thing we haven't touched: action verification. Say the agent sends an email on behalf of the user. How do you close the loop — confirming it actually sent, content matched intent, recipient was right — without forcing the user to manually re-check every action?

0 reply

0 recast

1 reaction