The 'pressure-test my thinking' prompt
When you've drafted something important — a strategy memo, a board update, a difficult email, a proposal — this prompt asks AI to be your toughest reviewer. Not a yes-machine. A skeptic.
Text Rosey · Schedule a call →The prompt
Copy and paste into ChatGPT, Claude, Copilot, Gemini — works anywhere.
When to use it
Use it on anything you've drafted that's going to be read by someone whose pushback matters — a CEO, a board, a client, a regulator, a key hire. The prompt's value is forcing AI out of the default agreeable-assistant mode and into the skeptical-reviewer mode where it actually catches things.
The "be direct, don't hedge" closing is load-bearing. Without it, AI tends to soften criticism into compliments. With it, you get the kind of feedback you'd hope a senior colleague would give before you sent the thing — but at 11pm the night before.
Worked examples
- Strategy memos before sending to leadership — AI flags unsupported claims, missing competitive context, jumps in logic. You decide what to fix vs. defend.
- Difficult emails before hitting send — AI catches tone issues, missing acknowledgments, unclear asks. Saves the email you'd otherwise regret.
- Board decks before the meeting — AI flags slides where the data doesn't support the claim, where the recommendation is buried, where the question the board will ask isn't pre-answered.
- Proposals to clients — AI surfaces the objections you haven't addressed and the price points you haven't justified.
- Performance review write-ups before the conversation — AI catches feedback that's vague, evidence that's thin, language that could land wrong.
- Code review write-ups before submitting — AI flags reviewer-friendly things you forgot (rationale, alternatives considered, test coverage of the change).
Why this prompt works
This prompt is the inverse of the default AI use pattern, and that's why it works. Most AI use in 2026 is asking AI to produce things faster — drafting, summarizing, generating. That has a ceiling on value. The high-leverage move is asking AI to evaluate and pressure-test things you've already produced.
McKinsey 2025 found AI high performers are nearly 3x more likely to have fundamentally redesigned workflows — and one of the most consistent redesigns is moving AI from "drafter" to "reviewer." The pressure-test prompt is that redesign at the individual level.
SHRM 2026 found 73% of directors and above report creativity improvements from AI vs. 65% of individual contributors. Directors are using AI for the kind of thinking work this prompt enables — pressure-testing, second-guessing, alternative perspectives. The prompt closes the gap by giving everyone the same access pattern.
Common questions
- Will the AI actually be tough, or will it just rephrase compliments?
- Depends on the prompt and the model. The 'be direct, don't hedge' closing matters. If you still get compliment-speak back, follow up with 'be more direct — what's the one thing that would most embarrass me about this draft?' That second pass usually surfaces the real issue.
- Does this work better with some AI tools than others?
- Marginal differences. ChatGPT, Claude, Copilot, Gemini — all do this pattern reasonably well in 2026. The bigger variable is the model size — larger reasoning models surface more substantive critiques than fast-and-cheap models.
- Should I share the AI's critique with my collaborators?
- Sometimes useful, sometimes not. Sharing it raw can feel like delegating editorial judgment to AI; sharing the critiques you decided to act on (and why) shows your judgment. The latter is the version most senior reviewers respect.
- What if the critique is wrong?
- Treat it like any reviewer's feedback — sometimes right, sometimes off-base, always worth considering. The value isn't 'AI is right'; the value is being forced to defend your draft against pushback before you ship it.
- Can I use this on something I'm not the author of?
- Yes. 'Pressure-test this draft I'm reviewing' works the same way. Useful for evaluating proposals, RFP responses, or PRs from your team — gives you a structured set of questions to bring to the conversation.
Sources
- AI high performers are nearly 3x as likely as others to say their organizations have fundamentally redesigned individual workflows — The state of AI in 2025: Agents, innovation, and transformation, McKinsey & Company (QuantumBlack, AI by McKinsey), 2025
- About 6% of organizations qualify as 'AI high performers' — those attributing 5%+ EBIT impact to AI — The state of AI in 2025: Agents, innovation, and transformation, McKinsey & Company (QuantumBlack, AI by McKinsey), 2025
- 73% of directors and above report creativity improvements from AI vs. 65% of individual contributors — The State of AI in HR 2026, SHRM (Society for Human Resource Management), 2026
Related
Text Rosey to begin.
Rosey is our executive-assistant bot. Text the number below — she'll ask two questions, offer three calendar slots, and put a 30-minute call on Jim's calendar.
Text Rosey · Schedule a call →