How to Run a Pre-Mortem Using Five AI Models

Pre-mortems are a powerful technique for anticipating risks and failure points before launching a project, product, or campaign. Traditionally, this process Great site depends on human expertise, brainstorming, and experience. But with advances in AI, you can significantly supercharge your pre-mortem workflow by harnessing multiple AI models. In this post, we'll walk through how to run a rigorous AI-powered pre-mortem using five different AI models—leveraging tools like Nick Launches and Suprmind—to perform decision intelligence, risk checks, and blind spot detection all within one conversation thread.

Why a Multi-Model Approach?

Individual AI models often have their own “blind spots” and are prone to hallucinations—generating confident but incorrect information. Just like in human decision making, having multiple viewpoints helps catch biases and reduces errors. A multi-model setup allows you to:

• Cross-check conclusions by comparing answers from different models.
• Surface disagreements to highlight potential blind spots you may not have considered.
• Integrate distinct strengths from various AI architectures for a more nuanced pre-mortem.
• Keep the entire risk-check process in a single conversational thread for easier referencing and exporting.

Tools Needed: Nick Launches and Suprmind

Both Nick Launches and Suprmind provide capabilities for running multi-model AI chats in one thread. This setup avoids scattered notes, tools hopping, and data loss. Their interfaces allow you to:

• Add multiple AI models into the same chat.
• Use customized prompts to trigger different types of analysis (e.g., risk checks, blind spot detection).
• Export the entire conversation into a structured format for sharing or archiving.

Step-by-Step Guide: Running Your AI-Powered Pre-Mortem

Here’s how you design and execute your pre-mortem session with five AI models. We’ll use the workflow structure popularized by decision intelligence professionals but enhanced with AI capabilities.

Step 1: Define the Project and Prompt All Models to Brainstorm Failures

Start by clearly describing your project or launch plan at a high level. Then, issue a pre-mortem prompt to all five models simultaneously:

"Imagine this project has just failed spectacularly. What are the top 5 reasons it could have failed? List them clearly."

Each AI model will independently generate failure reasons. Because the models vary in training data and approach, you’ll receive a rich spectrum of risk factors.

Step 2: Aggregate and Cross-Check Potential Failure Modes

Next, consolidate all failure reasons in a shared document or within the multi-model chat thread. Then, ask the models to:

• Identify overlapping failure types.
• Flag any unique but important risks mentioned by only one or two models.
• Rank the risks by likelihood or impact (if your prompt context allows).

This AI risk check step balances between majority consensus and minority warnings—you don't want to dismiss a rare but critical risk.

Step 3: Perform a Blind Spot Check through Model Disagreement

This is where the multi-model setup shines: ask the AI models to review each other’s failure reasons and highlight where they disagree or identify gaps. A prompt example might be:

"Based on other models' lists, are there any risks you believe were overlooked or underestimated? Please specify."

Disagreements here point to potential blind spots in your planning, exposing assumptions that need further scrutiny.

Step 4: Identify Mitigation Strategies

Once risks are settled, prompt each AI to recommend concrete mitigation steps, ideally tailored to the specific failure reasons they flagged. You may ask:

"For each key failure reason you identified, suggest practical actions the team can take to reduce risk or detect the issue early."

Comparing these across models gives you a breadth of possible interventions from different perspectives.

Step 5: Finalize and Export the Pre-Mortem Report

With Nick Launches or Suprmind, you can export the entire multi-model conversation into tidy, structured formats such as:

• Decision memos
• Risk registers
• Project briefings with AI-annotated risk assessments

I always ask, "What does export look like in practice?" because it's key that your AI insights become actionable artifacts, not just ephemeral chat logs.

Example Table: Summary of AI Failures & Mitigations

Failure Reason Number of Models That Identified It Mitigation Strategies Model Disagreement / Blind Spot Notes Inaccurate Market Research 5

• Conduct additional customer interviews
• Validate assumptions with pilot campaigns

Consensus across all models Technical Integration Failures 4

• Run end-to-end integration tests
• Allocate backup engineering resources

Model 3 underestimated risk severity Overestimating User Adoption 3

• Launch a beta with limited users
• A/B test onboarding flows

Raised as a blind spot by Model 5 Data Privacy Compliance Issues 2

• Consult legal early
• Implement privacy-by-design principles

Only addressed by Models 1 and 4 Poor Communication with Stakeholders 1

• Schedule regular update meetings
• Create transparent progress dashboards

Potential blind spot as only one model flagged it

Best Practices and Pitfalls to Avoid

Beware of Hallucination Moments

Multi-model chat setups can still produce confident but inaccurate or irrelevant data. Keep a running list of AI hallucinations encountered, and always validate critical risks with your real-world expertise.

Avoid Marketing Fluff and Enterprise Jargon

Prompt the AI for clear, step-by-step use cases and concrete examples rather than vague, high-level claims. This keeps your pre-mortem focused and actionable.

Don’t Expect AI to Fully “Solve” Decision Making

AI models are aids to help highlight blind spots and foster critical thinking, not oracles. Human judgment remains key to weighing tradeoffs, aligning mitigation with resource constraints, and making final calls.

Conclusion: AI-Enhanced Pre-Mortems for Smarter Launches

Running a pre-mortem with five AI models in one chat thread offers a powerful blend of decision intelligence, cross-model validation, and blind spot detection. With tools like Nick Launches and Suprmind orchestrating the conversation, you transform a standard brainstorming exercise into a structured and exportable risk intelligence asset.

Try integrating this approach in your next project kickoff. You’ll gain richer insights, catch errors earlier, and accelerate confident decisions—all while retaining full control over interpreting and applying AI-generated insights.

Have you tested multi-model pre-mortems or AI risk check workflows? Share your experience or tricky hallucination moments below!