Where you sit on the AI 0-7 scale

The levels are rough. Read them as one question: how far are you from purely manual coding?

0: isolation. 1: chat snippets. 2: copy-paste between LLM and IDE. 3: autocomplete. 4: AI IDE with an agent and review diffs. 5: CLI agents and your own workflows. 6: full-loop background agents. 7: custom orchestration for your own problems.

My personal mix right now is roughly: most of the week at 5, chunks at 6, rare experiments at 7.

That is my snapshot. It helps me see where I moved over a quarter.

When you put the scale in front of people, arguing about level names is almost useless. Sketch once where people sit today. Then you see what to buy and what to teach: agent rules, MCP, how to keep evidence. A slide that says “we use AI” does not help.