Chapter 36: Product Management

Teaching Aid

"The best PMs I've worked with share one quality: they have almost supernatural clarity about what they are building and why. Every feature spec is tight. Every roadmap decision is traceable to a user problem. Every stakeholder update tells the same story from a different angle. The rest of us — talented, well-meaning — spend half our time writing specs that confuse engineers, roadmaps that confuse executives, and research syntheses that confuse everyone. The gap is not intelligence. It is craft."

Product Management is one of the highest-leverage and highest-cognitive-load roles in any technology organisation. A single feature from ideation to launch requires discovery notes, a problem statement, user story mapping, a feature spec, a PRD, a sprint brief, acceptance criteria, a launch communication, a customer FAQ, and a post-launch retrospective. Most PMs either write some of these poorly or skip them entirely — and the downstream failures are predictable.

This chapter deploys two complementary plugins across the full PM workflow cycle. You will use 13 commands and 3 persistent agents to transform the bottleneck between PM judgment and the documentation that expresses it. The agent writes the first draft. You review, direct, and refine. The document reflects your judgment at the quality it deserves — in a fraction of the time.

🖥️ Fullscreen

What You'll Learn

By the end of this chapter, you will be able to:

• Diagnose the PM's cognitive load problem and explain why AI agents address the documentation bottleneck without replacing PM judgment
• Deploy two complementary plugins (official product-management + custom product-strategy) and configure them for your product using product.local.md
• Frame problems precisely using discovery briefs that separate the problem from the solution and the known from the unknown
• Design user research with interview guides grounded in five behavioural principles, then synthesise findings using thematic analysis and triangulation
• Write feature specs and PRDs that engineers can build from without ambiguity, with testable acceptance criteria and explicit scope boundaries
• Decompose PRDs into user stories with named personas, capability-focused "wants", and outcome-focused "so thats" — then prioritise the backlog using RICE with three mandatory challenge tests
• Plan sprints against real capacity, communicate status to three distinct audiences, and review product metrics using a North Star and L1 health indicator hierarchy
• Deploy three persistent agents (Research Intelligence, Stakeholder Update, Roadmap Coherence) that run your product intelligence layer continuously

Lesson Flow

Lesson	Title	Duration	What You'll Walk Away With
L01	The PM's Cognitive Load Problem	25 min	Understanding of the five PM roles and the documentation bottleneck AI agents address
L02	Plugin Architecture & Your Product Context	30 min	Both plugins installed, product.local.md configured, all 13 commands accessible
L03	Discovery Briefs — Framing the Right Problem	40 min	Problem brief for your product's biggest challenge, with discovery questions and evidence
L04	User Research — Interviews & Synthesis	45 min	Interview guide grounded in five behavioural principles, plus a research synthesis from simulated notes
L05	Competitive Intelligence	35 min	Competitive brief with feature comparison, positioning analysis, and strategic implications
L06	Feature Specifications	45 min	Feature spec with the five required sections, testable ACs, and explicit scope boundary
L07	PRDs for Multi-Team Initiatives	45 min	Full PRD with 10 sections, status gates, failure threshold, and GTM requirements
L08	User Stories & Story Mapping	40 min	User stories with named personas, capability-focused wants, and no compound ACs
L09	Roadmap Planning & Communication	40 min	Q3 roadmap in Now/Next/Later format with dependency mapping and capacity allocation
L10	Backlog Prioritization Frameworks	45 min	RICE-scored backlog with three challenge tests and quarterly priority decision
L11	Sprint Planning & Capacity	35 min	Sprint plan with capacity table, P0/P1/P2 backlog, risks, and definition of done
L12	Stakeholder Communication	40 min	Three stakeholder update versions (executive, engineering, customer) from the same status data
L13	Metrics, OKRs & Product Analytics	45 min	North Star metric, 5 L1 health indicators, OKR set, and monthly metrics review
L14	Continuous Intelligence — Agents & Retrospectives	75 min	Sprint retrospective with four questions, plus 3 persistent agents configured
L15	Chapter Summary & Quick Reference	20 min	All commands, all agents, key quality rules, and the PM workflow cycle diagram

Chapter Contract

By the end of this chapter, you should be able to answer these five questions:

1. What are the five roles a PM plays simultaneously, and how does the two-plugin architecture address the documentation bottleneck in each one?
2. How does product.local.md transform generic AI outputs into documents that sound like they were written by someone who has worked in your product team for two years?
3. What are the five sections every feature spec must contain, and why is the scope boundary (what is NOT included) as important as the solution description?
4. How do the three mandatory challenges (Strategic Override Test, Data Gap Test, "What Would We Regret?" Test) catch the errors that RICE scoring reliably misses?
5. How do the three persistent agents (Research Intelligence, Stakeholder Update, Roadmap Coherence) work together to maintain continuous product intelligence without requiring daily PM attention?

Prerequisites: Cowork Access

This chapter requires Cowork (set up in Chapter 28) and two plugins.

1. Install the official Product Management plugin. In the Cowork sidebar: Customize -> Browse plugins -> find Product Management -> click Install.
2. Install the custom Product Strategy plugin. In the Cowork sidebar: Customize -> Browse plugins -> Personal -> click + -> Add marketplace from GitHub -> enter https://github.com/panaversity/agentfactory-business-plugins -> find Product Strategy -> click Install.
3. Connect a working folder for practice files, same as Chapter 28.
4. Configure product.local.md — detailed in Lesson 2.

After Chapter 36

When you finish this chapter, your perspective shifts:

1. You see PM documentation as a judgment amplifier, not a chore. Every spec, PRD, roadmap, and stakeholder update is a vehicle for your judgment — and the agent removes the bottleneck between what you know and what you can document. The quality of your PM artifacts is no longer constrained by your available hours.
2. You have a working two-plugin architecture. Discovery briefs, interview guides, feature specs, PRDs, user stories, roadmap updates, backlog prioritisation, sprint plans, stakeholder communications, metrics reviews, and retrospectives are all installed, configured, and deployable through 13 commands.
3. You understand the quality boundaries. The agent writes first drafts. It does not make product decisions, commit to delivery dates, approve scope changes, or override your judgment. These boundaries are encoded in every skill file's "NEVER DO THESE" section.
4. You can extend. The interview guide framework transfers to any research domain. The spec quality rules apply to any technical writing. The three persistent agents work wherever continuous monitoring creates value — the PM workflow is just the first deployment.

Start with Lesson 1: The PM's Cognitive Load Problem.