AI for Product Managers From Research to a Shipped AI Spec

Use Claude, Perplexity, NotebookLM, Lovable, v0 and n8n to research, spec, prototype and evaluate real AI features.
What you'll learn

---

Draft a PRD with AI and edit it to a standard you would put your name on
Run cited market research and competitive teardowns in Perplexity, and audit the sources
Synthesise your own documents and interview transcripts in NotebookLM, anchored to exact passages
Turn user interviews into themes with a verbatim quote behind every theme
Build a working, clickable prototype with Lovable and v0, without writing code
Ask your product data questions in plain English and catch a confidently wrong answer
Prioritise a backlog with RICE and show how one confidence number changes the ranking
Write an AI PRD covering grounding, cost, failure modes and the human in the loop
Build a golden dataset and an eval rubric that captures your own quality bar
Read a trace and explain what the model actually did, step by step
Design a human-in-the-loop agent workflow and build a mini version in n8n
Present an AI feature to leadership as one page and five minutes you can defend
Requirements

---

Some product experience. The course assumes you already know what a PRD, a backlog and a roadmap are.
No coding, no maths, and no machine-learning background required.
Free accounts for Claude or ChatGPT, Perplexity, NotebookLM, Lovable and v0. Lessons are taught free-tier-first; some tools have usage limits on their free tier.
A Google account you are happy to use for a test calendar, for the Module 8 agent build.
Time for the capstones: roughly 8 hours for the core capstone and 4 for the advanced one, on your own schedule.
Description

---

This course contains the use of artificial intelligence.
You are being asked to "do something with AI." This course is what turns that into work you can show.
Most AI training for product managers is either a conference talk about how everything is changing, or a list of prompts. Neither one survives contact with a real backlog. This course is hands-on and tool-first: you research, spec, prototype, measure and defend one product, on camera, start to finish and every module ends with an artifact you keep.
By the end of this course you will have built
A one-page AI literacy cheat sheet in your own words
A PRD drafted with AI and edited to a standard you'd put your name on
A research brief synthesised from cited market sources and your own documents
A working, clickable prototype you can send to a user
A prioritisation doc where you can explain every number
An AI feature spec that covers grounding, cost, failure modes and the human in the loop
An eval rubric for a real feature
A mini agent workflow, wired and running
A capstone project: an AI product concept, evidenced, specced and prototyped
An advanced capstone: one page and five minutes you can take to leadership
This is a build course, not a survey course.
Everything is taught against one running product - Slotline, a B2B team-scheduling tool, so the work compounds. The PRD you draft in Module 2 is the thing you prioritise in Module 5, spec for AI in Module 6, write evals for in Module 7 and defend in the advanced capstone. You watch real sessions: a PRD drafted and then interrogated, a RICE table that re-sorts when one confidence number drops from 80% to 50%, an analytics answer that looks right and isn't, an agent that hits a calendar it can't read and has to stop. Nothing is staged after the fact, and where the AI got it wrong the wrong answer stays in.
The tools are the 2026 working set: Claude and ChatGPT for daily PM work, Perplexity for cited research, NotebookLM for your own corpus, Lovable and v0 for prototyping, Amplitude for plain-English analytics, and n8n for the agent build. You are taught free-tier-first where a paid tier adds something, it is named as an upside, not a requirement. No coding. No maths. No machine-learning background.
Who this course is for

---

Product managers, product owners and business analysts with some experience - the course assumes you know what a PRD, a backlog and a roadmap are, and spends its time on what AI changes about them.
It suits a PM who has been handed an AI initiative and wants a defensible way to run it, a senior PM who needs to cost and defend an AI feature in front of leadership, and an associate or aspiring PM who wants real work samples rather than a certificate.
Homepage

---