The Four Questions

The four questions are what to ask before opening any AI tool, because most “AI didn’t work” verdicts are really wrong-workplace verdicts: (1) What’s the time horizon — minutes, weeks, or recurring forever? (2) What needs to persist — nothing, a conversation, files, a whole project? (3) What needs to happen outside this session — file changes, system actions, scheduled runs? (4) Who needs to be in the loop — just you, or governed enterprise? The answers route the work to a chat, an in-context assistant, a coding agent, a workflow platform, or the mode of your enterprise suite.

Why it matters: People blame the model or the prompt when the work was simply in the wrong place — a multi-month project run in a chat, a one-shot question dragged into a coding agent. Same model, different workplace, opposite outcome. The 30-second version handles most decisions: question or project? Does anything need saving? Will I do this again?