CRIT Prompting Framework
Turn AI into a strategic thought partner through context and guided questioning
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 99%
CRIT structures an AI conversation around four elements: context, role, interview, and task. The user first explains the situation in detail, including relevant constraints, history, and desired outcomes. The AI is then assigned a vivid expert role suited to the problem. Instead of immediately generating an answer, it interviews the user with three to five questions, asked one at a time, to uncover missing information and challenge assumptions. Only after that discovery process does it complete the stated task. The mechanism works because it combines the user's specific knowledge with the model's broad pattern recognition. The resulting answer remains a first draft: the user must apply judgment, identify weaknesses, and continue refining it through conversation.
Origin
Geoff Woods developed CRIT after realizing that his strength was asking leaders questions, yet he was using ChatGPT merely to write emails and answer his questions. He reversed the interaction so AI would ask him questions that improved his strategic thinking.
Core principles
- 01Give AI more context than feels necessary
- 02Assign a vivid and relevant expert role
- 03Make AI ask questions before producing an answer
- 04Define a concrete task tied to a meaningful outcome
- 05Treat the first output as a draft rather than a verdict
How to run it
- 1
Supply context
Explain the problem, history, constraints, stakeholders, attempted solutions, and desired outcome in detail. Include information the AI could not know about you or the situation.
- 2
Assign a role
Describe the precise expert you want the AI to simulate. Include relevant domain expertise, industry knowledge, perspective, and behavioral characteristics.
- 3
Require an interview
Tell the AI to ask three to five questions, one at a time, before completing the task. Answer each question with enough detail to expose assumptions and hidden constraints.
- 4
Define the task
State exactly what the AI should produce after the interview, including format, audience, quality criteria, and desired outcome.
- 5
Refine the draft
Evaluate the response, state what you like and dislike, and specify the most important changes. Repeat until the output meets your judgment.
In the wild
A manufacturing CEO faced bankruptcy because a Japanese equipment supplier refused to restructure debt. Woods used CRIT to assign AI the role of an investment banker with restructuring expertise. After interviewing the CEO, it proposed a culturally informed strategy in which respected Japanese contacts could acquire the debt while allowing the original board to save face.
→ The proposal restored hope and initiated a restructuring process that the CEO later reported was moving forward.
An executive team used CRIT to have AI interview them about each director and produce personality profiles. Those profiles became a simulated board that reviewed a presentation, anticipated objections, and identified slides likely to derail the meeting.
→ The chairman described the next meeting and deck as the best the board had experienced.
Common mistakes
Using AI like Google
Asking a single short question encourages a broad answer that lacks the user's private context and strategic constraints.
Assigning a vague role
Calling the AI merely a marketing or business expert provides much less direction than specifying industry, specialty, audience, and perspective.
Trusting the first output
The initial answer should be judged and revised, not treated as an authoritative final result.
Is it for you?
Best for
It is best for complex decisions, strategic problems, ambitious goals, and tasks where important context resides in the user's head.
Not ideal for
It is excessive for simple factual lookups or routine tasks with a fully specified answer.
From the transcript
“So crit is just a really effective way to talk to AI. It stands for context, role, interview, task.”
“Give it lots of context, assign it a role where you tell it the kind of expert you want it to be, but then make…”
“The most important part is the interview. That's a game changer.”
From the episode
How to Use AI to Make Money, Grow Your Business, and Be More Productive