Python for AI/ML
Enough Python to work confidently with engineers.
Setting up Python environment — Jupyter, VS Code basicsTOOL
Python basics & data types — reading and adapting codeTHEORY
Data handling with Pandas — understanding how AI uses dataPRACTICAL
Data visualisation — Matplotlib, Seaborn for product metricsTOOL
REST APIs with FastAPI — understanding what your engineers are buildingTHEORY
Building apps with Streamlit — rapid product prototypingPRACTICAL
🎯 Project You'll Build
Product Metrics Dashboard — Load a product analytics dataset using Pandas and build a Streamlit dashboard showing retention, usage funnel, and feature adoption trends.
Tools: Python · Pandas · Streamlit
AI Overview
Shift from feature thinking to capability thinking.
AI vs ML vs Deep Learning vs GenAI — clear distinctions for PMsTHEORY
Enterprise AI maturity model — where your org standsTHEORY
AI transformation lifecycle — how AI initiatives get funded and shippedTHEORY
Ethics & responsible AI — bias, privacy, auditability for PMsTHEORY
Top AI use case archetypes — Classify, Summarise, Recommend, Predict, GeneratePRACTICAL
🎯 Project You'll Build
AI Use Case Portfolio — Map 3 products or features you know well. Classify each by AI type (Predictive / Generative / Agentic). Identify where AI could enhance or replace existing functionality.
Tools: ChatGPT · Claude
AI Coding
Cursor / Claude Code — build prototypes, write better PRDs.
Cursor & Claude Code setup — how to use AI coding assistants as a PMTOOL
Instruction files & spec-driven development — writing specs engineers can actually build fromPRACTICAL
Prompting patterns for coding — getting AI to prototype features quicklyPRACTICAL
Debugging with AI — understanding what went wrong in your AI featurePRACTICAL
Building simple tools without deep coding knowledge — validating product ideas fastPRACTICAL
🎯 Project You'll Build
AI-Driven PRD Generator — Use Claude Code to build a tool that takes a product brief and outputs a structured PRD: user stories, acceptance criteria, and AI capability requirements.
Tools: Cursor · Claude Code
GenAI Foundation
LLMs for PMs — understand what you're building with.
LLM overview & reasoning models — how they work and why it matters for productTHEORY
Prompt engineering patterns — getting reliable, structured outputsPRACTICAL
Chain-of-thought & few-shot prompting — advanced techniques for product use casesPRACTICAL
LangChain framework — building simple AI pipelinesTOOL
Building a user research summariser — automating insight extractionPRACTICAL
Evaluation & optimisation of LLM outputs — when to trust, when to questionPRACTICAL
🎯 Project You'll Build
User Research Summariser — Build an AI assistant that ingests user research transcripts and extracts structured themes: top pain points, feature requests, and satisfaction signals.
Tools: LangChain · Claude API · OpenAI API
Retrieval-Augmented Generation (RAG)
Product knowledge systems that actually know your product.
RAG architecture & overview — why AI hallucinations happen and how RAG prevents themTHEORY
Vector embeddings & chunking — preparing product docs for AI searchTOOL
Vector databases: ChromaDB / Pinecone — indexing product knowledgeTOOL
Indexing, retrieval & reranking — surfacing the most relevant product insightsPRACTICAL
RAGAS evaluation framework — measuring the quality of RAG outputsTOOL
LlamaIndex framework — building product knowledge Q&A systemsTOOL
🎯 Project You'll Build
Product Knowledge Base Q&A — RAG system over product documentation, user research, and competitor analysis. Query: 'What are the top user complaints about feature X?'
Tools: ChromaDB · Pinecone · LlamaIndex
AI Agents
Competitive intelligence and autonomous product research.
Agent architecture: LLM + Tools + Memory + PlanningTHEORY
LangGraph framework — building multi-step product research agentsTOOL
Agentic workflows — automating competitive analysis and user researchPRACTICAL
Short & long-term memory — agents that remember product contextTHEORY
Context engineering — keeping agents focused on product objectivesPRACTICAL
How to evaluate an agent's outputs as a PM — defining done for autonomous workPRACTICAL
🎯 Project You'll Build
Competitive Intelligence Agent — Reads a market brief → searches for competitive product insights → summarises findings by feature area → produces a structured competitive analysis report.
Tools: LangGraph · Claude API
AI Evals / LLMOps
Defining 'done' for AI features — the PM's most important skill.
AI evaluation frameworks — how to measure AI product qualityTHEORY
LLM-as-a-Judge patterns — using AI to score AI outputsTOOL
Evaluation process & product quality metrics — accuracy, trust, adoptionPRACTICAL
Observability & monitoring — what PMs need to watch in productionTOOL
Production quality gates — go/no-go criteria for AI featuresPRACTICAL
🎯 Project You'll Build
AI Feature Evaluation Framework — Define eval criteria for an AI product feature: accuracy, user trust, and task completion rate. Produce a quality scorecard with a go/no-go recommendation.
Tools: LangSmith · Claude API
Advanced Learning / AI Workflow Automation
Automate product ops and feedback loops.
AI task orchestration with n8n / Zapier / Make — visual workflow buildersTOOL
Trigger-based automation — event-driven product workflowsPRACTICAL
Integrating AI into product workflows — feedback triage, research synthesisPRACTICAL
MCPs in product development — connecting AI to Jira, Figma, analytics toolsTOOL
End-to-end workflow construction — automating the weekly product review cyclePRACTICAL
🎯 Project You'll Build
Product Feedback Triage Workflow — n8n workflow: user feedback → AI classifies by sentiment and topic → routes to relevant squad → auto-generates weekly 'voice of customer' summary.
Tools: n8n · Zapier / Make
Capstone — Foundation
Build your first AI product end to end.
Problem framing & system architecture designPRACTICAL
Tech stack selection — choosing the right AI components as a PMPRACTICAL
Core build sprint — implementing the AI product prototypePRACTICAL
Integration & evaluation — validating the AI product qualityPRACTICAL
Demo day & retrospective — presenting to a live review panelPRACTICAL
🎯 Project You'll Build
AI Product Blueprint — Define an AI-powered product feature from scratch: problem statement, AI use case selection, data requirements, success metrics, human-in-loop design, and system architecture sketch.
Tools: Full stack · Miro / FigJam
AI-First Product Thinking
Shift from feature thinking to capability thinking.
AI landscape overview: ML, GenAI, Predictive AITHEORY
What AI can and cannot reliably doTHEORY
Deterministic vs. probabilistic systems — why this matters for PMsTHEORY
From feature roadmaps to capability roadmapsPRACTICAL
🎯 Project You'll Build
Capability Mapping Exercise — Reimagine one existing product feature through AI capability lenses. Present before vs. after: current state, AI-augmented state, new risks, data requirements.
Tools: Miro / FigJam · Claude
Design Thinking for AI Products
Designing for ambiguity and uncertainty.
Designing for ambiguity & uncertainty — when the output is probabilisticTHEORY
Transparent AI interactions — when to show confidence, when to hide itTHEORY
Trust-building UI patterns — progressive disclosure, human overridePRACTICAL
Explainability & feedback loops — letting users correct the AIPRACTICAL
Responsible AI: bias, privacy, auditabilityTHEORY
🎯 Project You'll Build
Trust-Centred Feature Redesign — Pick a real AI feature with trust or opacity issues. Redesign with: confidence indicators, human override button, graceful degradation, feedback mechanism.
Tools: Miro / FigJam
Identifying High-Impact AI Use Cases
The most actionable PM skill.
AI use case archetypes: Classify, Summarise, Recommend, Predict, GenerateTHEORY
Data readiness assessment — evaluating feasibility before committingPRACTICAL
Value quantification methods — calculating ROI for AI use casesPRACTICAL
Build vs. Buy vs. Partner framework — the decision every AI PM must masterTHEORY
Writing AI-specific PRD sections — data requirements, edge case definitions, fallback logicPRACTICAL
🎯 Project You'll Build
Use Case Scoring Sprint — Score and prioritise 5 AI opportunities. Complete data readiness scorecard. Calculate ROI for the top use case. Produce a sequenced 3-use-case roadmap with AI PRD for #1.
Tools: Claude · Miro / FigJam
Automation vs. Human-in-Loop Strategy
Designing optimal human-AI collaboration.
Levels of automation spectrum — from assist to autonomousTHEORY
Automation bias risks — case studies of over-trust failuresTHEORY
Confidence thresholds — when to escalate to a humanPRACTICAL
Progressive automation strategy — building trust before removing humansPRACTICAL
🎯 Project You'll Build
Human-AI Collaboration Design — Design a human-in-loop workflow for a chosen product process. Map automation levels, confidence thresholds for escalation, fallback UX, and override audit trails.
Tools: Miro / FigJam · Claude
Metrics for AI Products
Defining success for AI systems.
Model performance metrics: accuracy, precision, recall, F1, latencyTHEORY
Product & business metrics: adoption, task completion, AI vs. baselinePRACTICAL
Trust metrics: override rates, NPS, complaint taxonomyTOOL
User corrections as a feedback signal — designing for continuous improvementPRACTICAL
🎯 Project You'll Build
AI Product KPI Dashboard — Define 6–8 KPIs across all three metric tiers (model / business / trust) for a chosen AI use case. Build in Power BI or Looker Studio. Present with go-live readiness assessment.
Tools: Power BI · Looker Studio
Managing Stakeholders in AI Initiatives
How to lead AI without losing the room.
Translating AI complexity into business language — for execs, legal, and financePRACTICAL
Risk communication strategies — how to present AI risks without killing the projectTHEORY
Governance & compliance considerations — what your org needs in placeTHEORY
AI roadmap planning — sequencing, dependencies, and change managementPRACTICAL
Build vs. Buy vs. Partner decision framework — presenting your recommendationTHEORY
🎯 Project You'll Build
Stakeholder Alignment Plan — Complete plan for launching an AI feature: business case for execs, risk communication for legal/compliance, governance framework, and a 12-month AI roadmap.
Tools: Claude · Miro / FigJam