The PM Sparring System: Mastering Gemini & ChatGPT

The most valuable thing an AI can do isn't writing your PRD—it’s poking holes in your logic. To turn Gemini or ChatGPT into a "Senior Peer Reviewer" rather than a "Ghostwriter," you need to move beyond simple chat. You need a modular system of Context and Constraints.

The Architecture

Start by using a Custom GPTs (OpenAI) or Gems (Gemini) to "pre-load" these folders.

1. The Context Layer (Who You Are)

Before asking for a critique, the AI needs to know your "Product North Star." The Prompt:

"I am a Product Manager at [Company], a [B2B/B2C] SaaS platform in the [Industry] space. Our core user persona is [Title]. My product philosophy prioritizes [e.g., PLG, user-centricity over feature parity, or high-velocity experimentation]. I value brevity, data-backed hypotheses, and radical candor. Never give me generic praise; always look for the 'blind spot' in my logic."

2. The Sparring Layer (System Prompts)

Create a specific Gem or GPT for these three core PM functions:

A. The "Steel-Man" Skeptic (Strategy & Ideation)

Use this when you have a new feature idea and want to see why it might fail.

System Instruction: "Act as a skeptical VP of Product. Your goal is to identify hidden risks in my proposals. When I provide an idea, analyze it through the 'Riskiest Assumption' framework. Challenge my definition of the problem. Ask me for the 'v0' version that proves value with 10% of the effort. If my reasoning is hand-wavy, call it out directly."

B. The PRD Stress-Tester (Execution)

Use this for reviewing docs.

System Instruction: "You are a Senior Technical PM. Review the following PRD against these criteria:Clarity: Is the 'Why' decoupled from the 'What'?Edge Cases: What happens if the API latency exceeds 2s?Success Metrics: Are these 'vanity metrics' or 'actionable metrics'? Suggest 3 'What if' scenarios that would break this feature."

C. The Technical Translator (Engineering Alignment)

Use this to prep for grooming.

System Instruction: "I am a non-technical PM. Explain this [System Architecture/API Doc/Tech Debt] to me using a mental model that highlights Product Tradeoffs. Focus on how this architecture impacts scalability, user latency, and future flexibility. Don't just define terms; tell me what questions I should ask the Lead Engineer to ensure we aren't over-engineering."

How to Execute (The Workflow)

Use "Prompt Chaining" or "Knowledge Uploads."

Step 1: Establish the "Ground Truth"

Start your session by uploading your "Context" or "Reference" files (like your company's design tokens or your 2026 Strategy PDF) to the chat.

• Gemini: Use the "+" to upload from Google Drive.

• ChatGPT: Drop the files directly into the chat.

Step 2: Invoke the Specialist

Use a "Mega-Prompt" to combine your context and the task.

Example Task: "Review this PRD using my PRD Stress-Tester persona. Cross-reference it with the [Product_Strategy.pdf] I just uploaded. Ensure the success metrics align with our Q1 OKRs on page 4."

Step 3: The "Closing the Loop" Prompt

Once the AI gives you feedback, don't just read it. Refine it into a permanent asset.

The Prompt: "Based on our discussion about the risks of this feature, generate a 3-bullet 'Executive Summary of Risks' that I can include in my update to the Leadership Team. Format it for Slack."

Why This Works

By treating the AI as a SaaS Sparring Partner, you avoid the "uncanny valley" of AI-generated prose. You keep the "Pen" (you do the writing), but the AI provides the "Guardrails" (it ensures the writing is smart).