First 5 AI use cases for a product manager

Five product manager tasks that are the easiest place to start with AI: each with a ready prompt and a hint on what to check in the answer.


How to use this

Open Claude Code or Codex (or a regular chat), copy the prompt and replace the text in square brackets with your own. Read the answer critically and refine the prompt if needed.

Data: in all examples use anonymized or test data, not real company documents. How to decide what you may give AI is covered in the AI Champion Blueprint. Check every AI answer: it can invent facts, which is why each scenario has a "what to check" line.


1. Map of hypotheses and risks

What AI does. AI turns an idea into testable hypotheses and finds the riskiest assumption.

Prompt:

Here is my idea in raw words:
[describe the idea]
Write 5 testable hypotheses using this template: "I think [segment] suffers from [pain], currently solves it by [way], and it hurts enough to [action or payment]". For each, name the riskiest assumption and how to test it within a week.

What to check. These are hypotheses, not facts. Test them on real people, not on the AI's opinion.


2. Competitor teardown

What AI does. AI builds a comparison table from the sources you give it.

Prompt:

Here are descriptions or links for 3 competitors:
[paste]
Build a table: who they sell to, price, strengths, weaknesses, how we can differ. Mark separately what comes from the sources and what is your assumption.

What to check. Check prices and features against the competitors' sites: data goes stale and AI may be wrong.


3. Product metrics

What AI does. AI proposes a main metric and supporting ones, and explains how to measure them.

Prompt:

My product: [description]. Propose one main metric (North Star) and 3-4 supporting ones. For each: how to calculate it, what data is needed, the pitfalls and how it can be gamed.

What to check. Do you actually have the data? Keep the metrics you can really measure.


4. PRD draft and a critic

What AI does. AI writes a draft PRD (product requirements document) and then criticizes it.

Prompt:

Write a PRD from my notes:
[paste notes]
Sections: problem, users, goals, non-goals, scenarios, metrics, risks. Then play a strict critic and find the 5 weakest points.

What to check. Is everything in the PRD taken from your notes? Remove what the AI added on its own.


5. Prioritizing ideas

What AI does. AI scores ideas with RICE and shows the assumptions behind every number.

Prompt:

Here are 8 product ideas:
[list]
Score each with RICE (reach, impact, confidence, effort) and show the assumption behind each number. Propose an order and say what is worth checking before trusting these scores.

What to check. The numbers are the AI's guesses. Put in your own data wherever you have it.

AI Sandbox Lab

Turn these scenarios into your own system

In the 6-session program you turn prompts like these into ready agents and templates for your role. Not sure the format fits? Book a free 15-minute call on Telegram.

See the program

← All materials

Ask a question