ai-product-manager-playbook

---

name: ai-pm-playbook description: "A comprehensive operating system for AI Product Management. Use this skill when planning, prototyping, evaluating, or launching AI-native products. It provides agentic workflows for roadmap planning under uncertainty, rapid prototyping, AI evaluations, cross-functional collaboration, go-to-market strategy, and responsible AI deployment."

---

AI PM Playbook

Overview

The ai-pm-playbook skill operationalizes the best practices of AI Product Management into executable, agentic workflows. It is designed to help product managers transition from traditional, process-heavy roles to the "builder mentality" required in the AI era.

This skill provides a structured approach to the entire AI product lifecycle, ensuring that products are built rapidly, evaluated rigorously, and deployed responsibly.

Use this skill when:

Prototyping a new AI feature or product.

Planning a product roadmap in a rapidly changing AI landscape.

Designing and running evaluations (Evals) for an AI model.

Structuring a cross-functional AI product team.

Developing a Go-To-Market (GTM) strategy for an AI product.

Implementing ethical guardrails and red teaming for responsible AI.

The AI PM Operating System

This skill is built on the premise that AI automates low-value PM tasks (like writing detailed PRDs) and elevates the need for strategic vision, judgment, and technical fluency. The workflows below are designed to augment these higher-order skills.

Core Workflows

Choose the appropriate workflow based on your current product development phase:

1. Prototyping and Rapid Experimentation

Move from static PRDs to interactive, "production-ready" prototypes.

Action: Decompose features, plan with AI, and build interactive prototypes.

Reference: See references/prototyping_workflow.md for the step-by-step guide.

2. Roadmap Planning Under Uncertainty

Shift from feature-based roadmaps to outcome-oriented planning.

Action: Define desired behaviors, use the Now/Next/Later framework, and apply the U.S.I.D.O. model.

Reference: See references/roadmap_uncertainty.md for the planning framework.

Template: Use templates/outcome_roadmap.md to structure your plan.

3. AI Evaluation and Metrics (Evals)

Move beyond basic accuracy to measure user experience, safety, and reliability.

Action: Define evaluator roles, supply context, set goals, and establish scoring rubrics.

Reference: See references/evaluation_metrics.md for the evaluation framework.

Template: Use templates/ai_eval_rubric.md to design your evals.

4. Cross-Functional Collaboration

Structure your team for success in the complex world of AI development.

Action: Implement a hybrid team structure, prioritize data readiness, and foster psychological safety.

Reference: See references/cross_functional.md for organizational best practices.

5. Go-To-Market Strategy and Trust

Launch AI products that meet evolving customer expectations and build trust.

Action: Define the 7 GTM pillars and prioritize transparency in data usage.

Reference: See references/gtm_strategy.md for the launch framework.

6. Ethics, Safety, and Responsible Deployment

Ensure your AI products are safe, trustworthy, and aligned with human values.

Action: Implement multi-layered guardrails and conduct rigorous red teaming.

Reference: See references/responsible_ai.md for the safety framework.

Template: Use templates/red_teaming_plan.md to structure your testing.

Self-Improving Loop

This skill incorporates a self-improving feedback loop to continuously refine your PM processes based on real-world execution data.

1. Collect Telemetry: After completing a major PM activity (e.g., a prototype sprint, an eval run, or a product launch), gather the outcomes, friction points, and user feedback. 2. Run the Loop: Execute scripts/pm_feedback_loop.py with the collected data. 3. Analyze and Adapt: The script will analyze the systemic friction and suggest updates to your templates, workflows, or evaluation rubrics to improve future performance.

Resources

scripts/pm_feedback_loop.py: The engine for continuous improvement of PM processes.

references/: Detailed guides for each of the 6 core workflows.

templates/: Standardized formats for roadmaps, evals, and red teaming plans.