One of the most valuable activities in project planning is a "pre-mortem" or a risk brainstorming session. You can simulate this with an AI by asking it to act as a "red team"—a friendly adversary whose job is to find weaknesses in your plan. This becomes even more powerful when you prime the AI with data from your past projects.

The Strategy: Learning from Lessons Learned

Before you ask the AI to find risks in your new project, you first give it the "Lessons Learned" or post-mortem document from a previous, similar project. This focuses the AI's analysis, encouraging it to look for recurring patterns or problems.

Step 1: Anonymize Your "Lessons Learned" Document Take the text from a previous project's retrospective. Remove all names and sensitive details. Focus on the sections that describe what went wrong or what could have been done better.

Example Anonymized "Lessons Learned":

"Key challenges from the [Previous Project Name]:

• The initial timeline did not account for developer onboarding to the new framework, causing a 2-week delay.
• The marketing team was not involved until the final month, leading to a rushed and less effective launch campaign.
• Scope creep was not managed effectively, with three major un-budgeted features being added mid-project."

Step 2: Craft Your "Red Team" Prompt Now, create a prompt that combines the context of your new project with the lessons from your last one.

Example Prompt:

"Act as a critical risk management expert. You are my 'red team' for a new project.

New Project Context: We are planning to build a new customer-facing analytics dashboard. The project timeline is 4 months.

Historical Context: First, please review the key challenges from a similar project we completed last year:

[Paste your anonymized "Lessons Learned" text here]

---

Based on both the new project context and the historical problems, please identify the top 5 potential risks for this new dashboard project. For each risk, briefly explain why it's a threat."

Why This Works

By providing historical context, you're helping the AI connect the dots. It won't just list generic risks. It will generate highly relevant potential issues, such as:

1. Risk of Schedule Delay: "Given the team's past issues with new technologies, the 4-month timeline for the new dashboard may not account for the learning curve of the new charting library."
2. Risk of Low Adoption: "The marketing team was engaged too late in the previous project. If they are not involved in planning for the new dashboard, we risk building a tool that no one knows how to use or promote."
3. Risk of Scope Creep: "The previous project suffered from unmanaged feature requests. The 'analytics dashboard' is a broad concept, and without a tightly defined MVP, it is highly susceptible to scope creep."

This AI-driven pre-mortem gives you a prioritized list of real, contextual risks, allowing you to build mitigation strategies into your project plan from day one.