Skill Shot · Strategic thinking · working
Frame an AI Use Case with 5W2H
By the end you can: frame a business AI use case by answering the relevant 5W2H questions and marking unanswered questions as unknown.
About 6 minutes with AryaPrerequisite: Draft: awaiting human review
Arya · AI teacher
Arya is waiting.
Scene 1 of 7 · 0:00 to 0:49
A technology label alone does not describe the work.
What is missing when someone says, “Let us use AI for contracts”?
Key points
- “Use AI for contracts” names a domain, not a task.
- A task description identifies work and its intended purpose.
- A clear handoff makes questions and gaps visible.
- Do not treat a possible application as a proven result.
Illustrative framing prompt
A vague proposal
- Proposal: use AI for contracts
- The specific task is not yet stated
- The intended user or reviewer is not yet stated
- No expected result is established
TakeawayDescribe the work AI supports, not only the technology involved.
Arya says (captions)
A colleague says, “Let us use AI for contracts.” That sounds like a direction, but it does not yet tell you what work the system would do. Would it read agreements, draft text, or flag something for a person to check? The phrase also leaves the purpose and the people involved unclear. Here is an analogy: a parcel marked only “office” is difficult to route; a useful handoff names what is inside and where it should go. The same applies to an AI proposal. A technology label is not a complete task description, and it is not evidence that the idea will work. We need a practical way to make the task and its context visible before assessing it.
Arya is an AI teacher and the voice is synthetic. This lesson is a draft and has not yet been reviewed by a person.
Ask Arya
Stuck, or want it said another way?
Arya is an AI. She answers only from this lesson, gives hints and not answers, and asks you to try it in your own words. Please do not share personal details.
Sign in and your questions are kept with your profile; as a visitor, nothing you ask is saved.
Check yourself
Three quick questions
1. The case says an algorithm flags problematic contract clauses for lawyer review. Which description preserves that division of work?
2. The source does not say when a particular LawGeex workflow reviews flagged clauses. What should you write?
Take it to work
This week, choose one task you are considering for AI. Draft a 5W2H statement, mark unsupported answers as unknown, and ask a colleague to identify the most important gap.
Try it, then ask Arya
Your practice note
Write a few sentences on how you would use this on a real problem from your work: what happened, what you did, and what you expect. Arya will read it against what this Skill Shot teaches. Please leave out names and personal details.
How useful was this Skill Shot?
One tap. Your rating tells us what to improve.
Remember
- Use 5W2H to make a task or problem clearer.
- Keep the stated task separate from the intended benefit.
- Name unknowns instead of inventing details.
- Frame AI as support for work when that is what the case describes.
Watch out for
- Describing AI as replacing the professional when the case describes AI assisting a lawyer.
- Treating an unanswered 5W2H prompt as permission to guess.
- Naming a technology without stating the task it supports.
Limits: 5W2H helps structure a task description; it does not prove that AI is suitable or that a use case will succeed.
Where this fits
- Skill set
- Strategy
- Capability module
- Frame business problems for AI
- Learning path
- New joiners and working professionals
Part 1 of 12
Next
- Assess whether an AI use case is suitableA clear task description is a useful starting point, but it does not establish suitability, controls, or likely results.
Full transcript and sources
1. A technology label alone does not describe the work. · 49s
A colleague says, “Let us use AI for contracts.” That sounds like a direction, but it does not yet tell you what work the system would do. Would it read agreements, draft text, or flag something for a person to check? The phrase also leaves the purpose and the people involved unclear. Here is an analogy: a parcel marked only “office” is difficult to route; a useful handoff names what is inside and where it should go. The same applies to an AI proposal. A technology label is not a complete task description, and it is not evidence that the idea will work. We need a practical way to make the task and its context visible before assessing it.
2. You can frame a use case by answering seven practical questions. · 48s
By the end, you will be able to frame a business AI use case with a set of practical questions. The observable result is a short description that states the task and purpose, identifies the people and context, and records the approach and scale where known. You will also distinguish what the case actually says from what you still need to ask. The questions are What, Why, Where, When, Who, How, and How much. Together, they are called 5W2H. You do not need to force a confident answer to every prompt. If the information is missing, write “unknown” and keep the gap available for discussion. That is more useful than presenting an assumption as a fact.
3. 5W2H turns a broad idea into questions about the work. · 50s
In plain language, 5W2H is a checklist of questions that helps you describe a task completely and clearly. The term is a compact name for the prompts: What, Why, Where, When, Who, How, and How much. What identifies the work; Why identifies its purpose. Where and When ask about setting and timing. Who asks which people or roles are involved. How describes the approach, and How much asks about amount or scale. The grid shows that a description can be clear about the task but still lack context, or the reverse. Aim to make both visible. This method structures the conversation; it does not tell you whether AI is appropriate, safe, or likely to succeed. Those questions need separate assessment.
4. Answer each prompt with evidence, then mark gaps as unknown. · 49s
Use the prompts in a steady order. First, write What: name the task, not just the technology. Second, write Why: use the purpose stated in the case, without promising an outcome that is not stated. Third, ask Where, When, and Who. These make the setting, timing, and people visible. Fourth, ask How and How much. Describe the approach and amount only when the information supports it. Finally, review each answer: is it stated, or are you inferring it? Label any unsupported detail as unknown. This last check matters because a neat-sounding description can hide assumptions. You can still share an incomplete frame; the point is to show colleagues exactly what is known and what they need to clarify.
5. The LawGeex example names a task, purpose, approach, and reviewer. · 54s
Now apply the prompts to the same case. The supplied transcript describes LawGeex as an Israeli startup using natural language processing to automate contract review. For What, state that the algorithm machine-reads legal agreements and detects problematic clauses. For Why, use the transcript’s stated purpose: improving the speed and quality of lawyers’ work. For Who, include both the algorithm and the lawyer who reviews flagged clauses. For How, say that the system machine-reads agreements and flags problematic clauses. The description does not establish a particular organisation’s setting, when the review occurs, or the amount handled in this workflow. Mark those details unknown rather than borrowing figures from another example. The non-example is saying “AI replaces the lawyer”; the transcript instead describes AI plus lawyer, with the lawyer reviewing flagged clauses.
6. A good next question targets the gap the team has identified. · 61s
Try a new, related problem from the supplied material. The transcript says algorithms can generate natural-language earnings reports from financial data, with tone and sentiment reflecting the financial results. For this challenge, treat What, Why, and How as already stated. The team has not specified where the generated report will be used. Choose the next question that directly clarifies that missing setting. You have one decision to make: which prompt best fills the gap? Pause for a moment and choose before reading the feedback. Remember that 5W2H is useful here because it makes a missing part visible; it does not supply the answer. You will need to ask the team rather than assume the destination from a different example.
7. A useful use-case frame makes facts and unknowns visible. · 51s
Keep the practical lesson simple. A broad idea becomes easier to discuss when you state the task, its purpose, the people and context, the approach, and the amount where known. Use What, Why, Where, When, Who, How, and How much as prompts, not as a form that demands invented answers. The LawGeex case shows why preserving roles matters: the described algorithm flags clauses, and a lawyer reviews them. This week, choose one task you are considering for AI. Draft a 5W2H statement, mark unsupported answers unknown, and ask a colleague to identify the most important gap. Remember that this frame does not prove the idea is suitable or establish its results. A useful next step is to assess suitability, risks, and controls separately.
Sources
- S1 · Test Admin, approved lesson notes; reference transcript: AI Applications for Growth, Module 1, supplied excerpt.
- S2 · 5W2H prompts and steps supplied in the author notes.
Attribution: Test Admin. Version 0.1.0.