In my continuing effort to make AI into a usable tool, I told it I don't care for connectors, because I don't trust that my context will be interpreted correctly. The response sorta surprised me: "Fair enough. That control-then-trust approach fits your line of work anyway. Scope defines what touches what."