THE DIRECT ANSWER

Before an AI tool starts real work, name who can stop it, what the team will do next, what must be saved, and how the team will learn.

The short answer

An AI tool needs an exit plan before it starts real work. The plan should say when to stop, who can stop it, how work will move, what to save, and what must happen before it starts again.

This is not a sign that the tool will fail. It is a sign that the team is ready to own the work.

Set the stop signs first

Do not wait for a bad day to choose the limit. Write the signs that should stop the tool. A sign may be bad data, a missed check, an odd result, or a task the tool should never touch.

Make each sign easy to see. A vague rule like “stop if it looks wrong” puts too much weight on the last person in the chain.

Give one person the power to act

A stop rule is weak if no one owns it. Name one role that can pause the tool at once. That person needs the right access and a clear way to tell the team.

The owner does not need to fix the whole system alone. The owner does need to make the first safe call.

Keep a safe path for the work

The work may still matter when the tool stops. Keep a plain backup path. Say who takes over, which steps go by hand, and what can wait.

Test this path before it is needed. A backup that no one has tried is only a hope.

Save the facts, then learn

Save enough to see what took place. Keep the input, output, time, owner, last change, and reason for the stop. Do not save private data you do not need.

Then set the restart rule. Name the fix, the test, and the proof the team needs. A fast restart is not the goal. A safe, clear restart is.

ORIGINAL WORKSHEET

The five-part AI exit plan

Fill this out before the tool starts real work. Keep the answers short, clear, and easy to find.

  1. 01 · StopWhat event should stop the tool?Name a clear sign, such as bad data, a missed check, or a result outside the set limit.
  2. 02 · OwnerWho can stop it right now?Name one role with the power, access, and duty to act.
  3. 03 · BackupHow will the work keep moving?Write the safe manual path, who takes over, and what work may wait.
  4. 04 · RecordWhat must the team save?Keep the input, output, time, owner, change, and reason for the stop.
  5. 05 · ReviewWhat must be true before the tool starts again?Set the fix, test, owner, and proof needed for a safe restart.

Original evidence: Tim Yslava’s five-part AI exit-plan worksheet. It is an operating guide, not a claim about a customer, product, or result.

Four takeaways

  • Set clear stop signs before real work starts.
  • Name one owner who can pause the tool.
  • Keep and test a safe manual path.
  • Save the key facts and require proof before restart.