Lesson Content

The pre-mortem technique.

"I am planning to [your plan]. Imagine it is 12 months from now and this plan has failed badly. Without being asked to stay positive, tell me the five to seven most realistic reasons this failure might have occurred. Be specific – not generic risks, but the specific failure modes most likely given what I have told you."

"Without being asked to stay positive" is essential – without it, ChatGPT often balances each critique with encouragement, diluting the analytical value.

Devil's advocate.

"I believe [your position]. Argue the strongest version of the opposite position. Do not agree with me at the end – I want a genuine critique of my position, not a 'but on the other hand' conclusion. Find the weakest points in my reasoning and attack them."

"Do not agree with me at the end" overrides ChatGPT's default to balanced, diplomatically framed responses. For devil's advocate to be useful, you need a genuine challenge.

Responding productively to challenge.

• Curiosity over defensiveness: "What is true about this critique that I have not wanted to acknowledge?"
• Distinguish applicable from inapplicable critiques: engage with the ones that reveal real weaknesses
• Update when warranted: if a critique is compelling, change your plan

The important limitation.

ChatGPT constructs critique from general knowledge and your description – not from knowledge of your industry specifics, your team's actual capabilities, or your real market. Supplement ChatGPT's analysis with critique from people who have direct knowledge of your context.