Use-Case Fit Worksheet

Use this worksheet before turning a workplace task into a prompt or automation idea. It helps you decide whether AI should support the work, which role it should play, and what review guardrails are needed.

This is the hands-on companion to Deciding Where AI Fits. Use the lesson for the reasoning, then use this worksheet to make the call on one real task.

Task Snapshot

Write one sentence for each item:

• Task: What work do you want help with?
• User: Who will use the output?
• Source material: What facts, notes, documents, or examples can you provide?
• Output: What should come back from the AI tool?
• Review owner: Who is responsible for checking the result?

If you cannot fill in these five lines, the task is not ready. Clarify the workflow before prompting.

Fit Test

Mark each statement as yes, partial, or no:

• The task can be described clearly.
• The model can be given enough context to work from.
• The output can be checked by a person.
• The output is a draft, structure, summary, recommendation, or question list.
• A wrong answer would be caught before it affects another person, system, or commitment.

Scoring guidance:

• Mostly yes: strong fit.
• Mix of yes and partial: partial fit; narrow the AI role and keep review close.
• Several no answers: weak fit; do not use AI until the task is narrowed or better context is available.

Role Selection

Choose one primary role:

• Drafter: create a first version for a person to revise.
• Organizer: turn messy input into clearer sections, groups, or steps.
• Reviewer: compare work against known criteria.
• Sparring partner: suggest alternatives, questions, risks, or tradeoffs.
• Extractor: pull specific details from provided source material.

Write the role in the prompt. For example:

Act as a reviewer. Compare this draft against the checklist below. Do not rewrite it yet. Return only missing requirements, unclear claims, and suggested questions for human review.

Guardrail Gate

Before using AI, confirm these guardrails:

• Sensitive details are removed, generalized, or explicitly approved for use.
• The prompt says what the model should not invent.
• Important claims must be labeled as grounded, unsupported, or needing verification.
• The output will be reviewed before it is shared, acted on, or used to update a system.
• Any high-consequence decision remains with an accountable person.

If any guardrail is missing, add it before running the prompt.

Decision Line

Complete this sentence:

This task is a [strong / partial / weak] AI fit. AI should act as [role]. The output can be used only after [review step].

Examples:

• This task is a strong AI fit. AI should act as organizer. The output can be used only after the meeting owner verifies decisions and open questions.
• This task is a partial AI fit. AI should act as sparring partner. The output can be used only after a person checks the proposed tradeoffs against the actual constraints.
• This task is a weak AI fit. AI should not draft the final answer. A person should first provide source material and decision criteria.

Worked Example

Scenario: summarize meeting notes into decisions and open questions.

Task Snapshot

• Task: Turn the meeting notes into a short list of decisions, open questions, and follow-up tasks.
• User: The meeting owner will use the output to prepare a recap.
• Source material: The notes from the meeting, including attendee comments and any stated owners or due dates.
• Output: A table with decision, owner, due date, source sentence, and review note.
• Review owner: The meeting owner checks the output before it is shared.

Fit Test

• The task can be described clearly: yes.
• The model can be given enough context to work from: yes, if the notes are included.
• The output can be checked by a person: yes, by comparing each item to the notes.
• The output is a draft, structure, summary, recommendation, or question list: yes, it is a summary and action list.
• A wrong answer would be caught before it affects another person, system, or commitment: yes, if the meeting owner reviews before sending.

Decision: strong fit, with review.

Role Selection

AI should act as extractor and organizer. It should pull details from the notes, group them into a usable shape, and avoid creating owners or dates that are not present.

Prompt role line:

Act as an extractor and organizer. Use only the meeting notes below. Do not invent owners, dates, decisions, or commitments. If something is missing, mark it as “Not stated.”

Guardrail Gate

• Sensitive details removed or approved: confirmed before notes are pasted.
• Unsupported details forbidden: included in the prompt.
• Important claims labeled: each item includes a source sentence.
• Review before sharing: meeting owner reviews the recap.
• High-consequence decisions remain with a person: yes; the recap does not create new commitments.

Decision Line

This task is a strong AI fit. AI should act as extractor and organizer. The output can be used only after the meeting owner checks each decision, owner, due date, and source sentence.

Quick Reference: Weak-Fit Warning Signs

Treat the task as weak or not ready yet if any of these are true:

• You cannot describe the task in one sentence.
• You cannot provide the source material the answer should rely on.
• No person can tell whether the output is correct.
• A wrong answer could reach a person, system, or commitment before review.
• The task asks AI to make the final decision instead of supporting one.
• The task includes sensitive or high-consequence material without an approved boundary.
• The output needs facts, owners, deadlines, or approval that nobody has provided.

If you see one of these signs, narrow the task. Ask AI for questions, a checklist, an assumption list, or a missing-information inventory instead of a final answer.