Run an AI pre-mortem: assume it failed, find out why — before it does
A pre-mortem flips the most useful switch in decision-making: instead of asking "will this work?", it assumes the plan has already failed and asks what killed it. The trick works because imagining a certain failure licenses people — and models — to name the risks that optimism keeps off the table.
Decidi runs the exercise properly. A Pre-Mortem Analyst leads a council that starts from "it is a year later and this failed" and works backwards to the most plausible causes — with a Devil’s Advocate pushing past the comfortable answers, a Risk Officer ranking the downside, and a Pragmatist testing which mitigations are actually doable. Because several independent frontier models run it in parallel, you get failure modes one mind would not generate. The verdict names the most likely cause of death, the assumptions it travels through, the mitigations in priority order and the early-warning signals — reviewed by a Final QA audit before you see it.
- The single most likely failure mode, named and ranked — not a vague risk list
- The chain of assumptions the failure travels through, made explicit
- Failure modes from several independent models, not one imagination
- Mitigations ranked by what actually prevents the failure, not what is easy
- Early-warning signals to watch, so you catch it starting
- A proceed / fix-first / stop call you can defend
Part of: Problems we solve
A pre-mortem report: the most probable cause of failure, the assumptions behind it, the mitigations in priority order, and the signals that tell you it has started.
Common questions
What is a pre-mortem and why does it work?
A pre-mortem imagines the plan has already failed and asks what caused it — the reverse of asking whether it will succeed. Assuming failure as a premise removes the social and psychological pressure to be positive, so risks people would soften or self-censor get named plainly. It regularly surfaces the thing that would have sunk the project.
Why use AI for a pre-mortem?
A team’s pre-mortem is limited by the team’s imagination and its politics — people hesitate to name the boss’s idea as the cause of death. Independent models have no such stake, and several of them generate a wider space of failure modes than one room does. Decidi structures it: generation, challenge, ranking, mitigation.
When should I run one?
Before commitment, while changing course is still cheap: before a launch, a big hire, a pivot, a major purchase, signing a lease or a funding round. The output is most valuable when there is still time to fix the top failure mode rather than merely brace for it.
How is a pre-mortem different from a red team?
A red team attacks your plan from outside, hunting for the ways an adversary or reality breaks it. A pre-mortem works backwards from an assumed failure, which is better at surfacing slow, internal causes — drift, dependency, the assumption everyone forgot was an assumption. Decidi offers both; they compound well.
Try it on your own decision
Put your question to a council of GPT, Claude, Gemini and Grok — they debate it, a Final QA audit reviews it, and you get one clear verdict. 1,500 free credits to start — no sign-up, no card required.
Start free