Map
Framework

Minimum viable judgement

The question it answers

Before handing a task to AI, which three questions must a manager answer first, so the task is not misplaced, under-briefed or blind to the damage if it goes wrong?

Origin
Anthropic, 2024; Anthropic, 2025; Dell'Acqua et al., 2023; Parasuraman et al., 2000
Diagram type
Process by phase
Level of analysis
Individual
Topics
AI for leaders
Glass sphere with a glowing neural-network core, icon for AI for leaders

Diagram

Minimum viable judgement

  1. Three questions before delegating
  2. Is the problem worth giving to AI?Is the task within AI’s reach, and does the trade in time and cost pay off?
  3. What does AI need to know to do it well?The smallest set of information that still gets it right: context, examples, constraints.
  4. What is the damage if it is wrong?Who bears it, how much, and who checks before the result goes out.

When to use it, and when not to

Use it when

Use it as a quick check before each new task you give to AI, when you do not yet need a whole process. Question 1 leads to The AI delegate matrix if you need to score it properly, question 2 to The five-part context bundle, question 3 to The five-rung verification ladder.

Avoid it when

The three questions are this site’s grouping of three more detailed frameworks; no source prescribes this set of three, and “minimum viable judgement” is the site’s own label. For work with large consequences, three questions do not replace the delegate matrix and the verification ladder.

How to apply it, step by step

  1. Worth giving: (Anthropic, 2024) advises finding the simplest solution and notes that AI systems often trade waiting time and cost for quality, so you should weigh when that is worth it; (Dell'Acqua et al., 2023) shows AI helping on tasks inside its reach and hurting on a task outside it.
  2. What it needs to know: (Anthropic, 2025) advises finding the smallest set of information that still lets the model do the right thing; small does not mean short.
  3. What if it is wrong: the cost of consequences is one criterion for choosing a level of automation ((Parasuraman et al., 2000)).

Pitfalls

Pitfall

Answering question 3 with “it will probably be fine”. In Mata v. Avianca the damage was a court filing with fake cases and a US$5,000 penalty ((Mata v. Avianca, Inc., 2023)).

Sources

References

Listed A–Z by author surname, Harvard style. A source with a DOI links to https://doi.org/…; one without links to its page or says plainly that it has no DOI.

  1. Anthropic (2024) Building effective AI agents. Anthropic Research.No DOIanthropic.comCited in 6 articles
  2. Anthropic (2025) Effective context engineering for AI agents. Anthropic Research.No DOIanthropic.comCited in 4 articles
  3. Dell'Acqua, F. et al. (2023) Navigating the jagged technological frontier: field experimental evidence of the effects of AI on knowledge worker productivity and quality. (Harvard Business School Working Paper No. 24-013).doi:10.2139/ssrn.4573321Cited in 2 articles
  4. Mata v. Avianca, Inc. (2023) 678 F. Supp. 3d 443 (S.D.N.Y. 2023), No. 22-cv-1461, Opinion and Order on Sanctions, 22 June 2023.No DOInhd.uscourts.govCited in 3 articles
  5. Parasuraman, R., Sheridan, T.B. and Wickens, C.D. (2000) ‘A model for types and levels of human interaction with automation’, IEEE Transactions on Systems, Man, and Cybernetics, Part A: Systems and Humans, 30(3), pp. 286–297.doi:10.1109/3468.844354Cited in 2 articles