Knowledge Tree
and Butterfly
More thinking, less grinding

Chatbot or do-bot?

Self-assessment · eight questions · two minutes

If AI so far has felt like something done to you, you have met the chatbot. A do-bot is the same model wired into files, a shell, tests and version control. It does the work and shows you what it did.

Eight questions place you on the ladder and say what the next rung costs. Answer for how you actually work, not how you would like to. Nothing is recorded until you choose to book.

Rung 1ChatbotYou paste in and copy out.
Rung 2SidekickAI in your editor. It suggests; you accept or reject.
Rung 3Do-botA harness with files, shell, tests, permissions. It shows you the diff.
Rung 4OrchestratorHandover files, unattended runs, agents with roles, a manifest.
  1. Q1Last time AI helped with real work, where did it happen?
  2. Q2When it changes something, how do you know what it did?
  3. Q3Your inputs change: a new data drop, a corrected figure, a client moves a parameter. What happens to the outputs?
  4. Q4Where does the project's memory live between sessions?
  5. Q5Version control.
  6. Q6Tests.
  7. Q7Could a colleague run your analysis on a clean machine tomorrow?
  8. Q8What can the AI do without asking you?

Your result

Answer the 8 questions

AI use–/12
Foundations–/12

The button opens an email to us, copied to Rabbie, our robot, who keeps the diary. Your result travels in it so we can prepare. You can change or delete it before you send, and nothing leaves your browser until you do. The self-assessment is written for people who work with data and evidence. If you build software, the mirror-image version is Could you speak statistician?.

Back to the homepage →

Score –/24result ↓