AI Made EZ: The Pre-Mortem Simulator

Of all the creative uses of AI, using it to intentionally "break" your ideas is one of the most valuable yet underutilized strategies. We typically use LLMs to build things—drafting plans, writing code, or generating ideas. But these models are equally skilled at playing the role of a "Red Team," simulating failure scenarios to help you disaster-proof your work before it ever sees the light of day.

An Example

I was recently preparing to launch a new internal policy change regarding remote work hours. I had drafted what I thought was a fair, balanced announcement. I felt good about it. But before hitting send, I decided to run a "Pre-Mortem." I pasted the announcement into the AI and asked it to adopt the persona of a skeptical, overworked employee who feels micromanaged. The AI’s response was brutal—and incredibly insightful. It immediately pointed out a vague phrase about "core hours" that could be interpreted as a stealth pay cut for hourly staff. It wasn't, but the perception would have been disastrous.

The Simulation Benefit

By asking the AI to simulate a negative reaction, I spotted a PR crisis before it happened. This is the "Devil's Advocate" benefit. Our own biases often blind us to how our work will be received by people with different incentives or contexts. An LLM has no such bias; it can instantly switch lenses to view your project from the perspective of a confused customer, a strict regulator, or a competitor looking for weakness. It turns the abstract anxiety of "what if this goes wrong?" into a concrete list of fixable issues.

Practical Uses of Simulation

This technique works across almost any domain. Sales teams use it to roleplay difficult prospects who have endless objections. Software architects use it to predict edge cases where a new system design might fail under load. Lawyers use it to anticipate counter-arguments from opposing counsel. Even job seekers use it to simulate tough interview questions based on their specific resume gaps.

The Prompt

Here is the prompt I used to stress-test my announcement:

"I am going to paste a draft of a new policy announcement. I want you to act as a skeptical, highly critical employee who is looking for reasons to be upset by this. Read the draft and tell me exactly which sentences sound unfair, vague, or prone to misinterpretation. Be brutal."