My rule of thumb is to turn your doubt into a question with experience attached to it. E.g. when someone says “there’s an API for that, it should be easy”, I note that not all APIs are equal (enumerate examples) and it must be reviewed before we commit to deadlines. And until that we must assume it’s not “easy”. Ask if research was done and whose responsibility it would be. If it wasn’t, then why we decided to meet. Hanging responsibilities onto dangling tongues explicitly makes them dangle less.

How hard to push back depends on the nature of discussion. Being too pessimistic in a research phase is not useful. But if it’s a contract worth half a year or more, spending few days evaluating assumptions is wise.

Remember that fast thinking is not actually fast, but shallow and/or optimistic.