AI Product Discovery
Knowing what to build before you build it
TL;DR
Knowing what to build before you build it
> Overview
AI makes building fast. Cursor, Claude Code, and Replit Agent mean a prototype can exist in hours. This is revolutionary, but it creates a new risk: building the wrong thing faster. The PM's highest-value skill is no longer managing a backlog. It is identifying which problems are genuinely worth solving with AI, and which are just impressive demos that nobody needs. This module teaches the discovery frameworks specific to AI products.
> Why This Matters for Your Product
Every team has more AI ideas than capacity. The teams that win are the ones that ruthlessly filter for ideas where AI creates genuine user value, not just technical novelty. Without a disciplined discovery process, you will build features that impress in demos but get ignored in production. This module is the filter.
> Interactive & tools
Building is cheap now. Discovery is what separates successful AI products from expensive demos.
Agent vs. automation vs. human
Decision Matrix: Agent vs. Automation vs. Human
Agent
- •Task requires judgment and patterns
- •Handles ambiguity, not pure rules
- •Best for: research, drafting, analysis
Best for: Judgment + pattern-based work
Automation
- •Rule-based, no ambiguity
- •Zapier, n8n, scripts
- •Best for: triggers, data sync, alerts
Best for: Deterministic workflows
Human
- •Empathy, creativity, legal accountability
- •Keep human with AI assistance
- •Best for: escalation, approval, nuance
Best for: High-stakes or subjective decisions
Case studies
Real-world examples
Related Engineering Patterns
These are the technical patterns your engineering team will implement. Understanding them helps you have better conversations.
Key Product Decisions
- [01]Which of your product's workflows score highest on the 5-dimension audit?
- [02]For your top AI candidate, does it pass the 80% automation test?
- [03]Does the user trust profile (stakes, reversibility, transparency) support an autonomous agent or a copilot?
- [04]Are competitors already shipping agent features in your category? Where are the gaps?
- [05]Is traditional automation (Zapier, scripts) a better fit than an AI agent for any of your candidates?
Ask Your Engineering Team
- →For our top workflow candidate, what is the technical feasibility to automate 80% of it?
- →What data do we have access to that would power the agent's decisions?
- →What is the simplest possible prototype we could build to test user reaction?
- →What existing tools or MCP servers could we leverage to accelerate the prototype?
Unlock the decision framework
Free account — no credit card required. Sign up to see the full decision checklist and the questions to ask your engineering team.
Sign Up FreePlay the interactive AI Product Discovery game
Practice the decisions from this module in an interactive game. Sign up free to play and save your progress.
Sign Up Free to PlaySee the full decision framework
Sign up free to see this module's Key Decisions, the questions to ask your engineering team, and play the interactive AI Product Discovery game.
Sign Up Free