Quality checklist

Workslop prevention checklist.

How to stop AI from creating polished-looking work that makes someone else clean up the mess.

Why this matters

Polished output is not the same as useful work.

AI can make weak thinking look finished. This is especially risky at work because the person receiving the output may spend more time fixing it than you saved making it.

Use today

Check usefulness before you send.

Before sharing AI-assisted work, make sure it has a clear purpose, correct context, visible assumptions, and an obvious next action.

The six workslop checks.

1

Purpose

Can the recipient tell what decision, action, or understanding this output is supposed to support?

2

Context

Does it include enough background to be useful, without dumping irrelevant detail?

3

Accuracy

Have you checked facts, names, numbers, dates, links, and claims before sharing?

4

Specificity

Could this have been written for any company, or does it actually fit your task?

5

Ownership

Are assumptions, open questions, and human judgement clearly marked?

6

Next action

Does the recipient know what to do next, or have you handed them a tidy-looking puzzle?

Copy-paste prompt: workslop check.

# ROLE You are my AI work quality reviewer. # TASK Check whether this AI-assisted draft is actually useful or just polished. # DRAFT [paste non-sensitive draft] # OUTPUT FORMAT 1. What is useful 2. What is vague or generic 3. What could create work for someone else 4. Facts or assumptions I need to check 5. Edits that make the next action clearer 6. Final ready-to-send version only if the draft is good enough # RULES - Be direct. - Do not make unsupported claims. - Mark anything that needs human verification.

Before you send AI-assisted work.

Green light

  • The purpose is obvious.
  • Facts have been checked.
  • The recipient has a clear next action.

Slow down

  • It sounds impressive but vague.
  • It contains claims you have not verified.
  • It hides important uncertainty.

Do not send yet

  • The recipient would need to redo the thinking.
  • It includes sensitive details unnecessarily.
  • You cannot explain how the output was checked.
Check an AI answer Repair a weak prompt Check safe use