StewAI Blog · Decision Science · 15 min read
The Premortem: Assume It Already Failed, Then Explain Why
A premortem beats ordinary risk brainstorming because it changes the tense: instead of asking what could go wrong, it declares the plan already failed and asks why. Decision science says that certainty surfaces about a third more reasons. We point M002 Project Premortem at a real launch plan and watch thirteen failure causes come out.
Frequently asked questions
What is a premortem, and how is it different from risk brainstorming?
A premortem, from Gary Klein, declares the project has already failed and asks why. That change of tense, called prospective hindsight, surfaces roughly a third more failure reasons than asking what could go wrong, because certainty about the outcome unlocks more specific explanations.
What did the M002 Project Premortem run actually produce?
Pointed at a real launch plan, it generated failure causes through five independent lenses, deduplicated and grounded them, and emitted a risk register of thirteen causes, each with a criticality score and a preventive mitigation, plus an overall viability verdict.
Is there an autonomous agent inside the recipe?
No. Every step is declared up front, the model reasons only inside steps against fixed schemas, the lenses run in isolation so they cannot anchor each other, and the scores and verdicts are computed and thresholded in code.
Read the full article on StewAI