Prompt Like a Product Manager: A Founder’s Guide to Using AI Well
Too many people prompt AI like they’re ordering lunch — quick and low-effort. They throw in a vague request like "summarize this" or "give me ideas," and wonder why the result feels generic, shallow, and unusable.
Product managers can do better, not by mastering “prompt engineering” as a new discipline, but by applying the skills they already use every day: framing problems, setting expectations, and asking the right questions from the right people.
That’s what this article is about: how to prompt like a real, AI-enabled product manager—using the same mindset you bring to discovery interviews, roadmap reviews, and half-baked feature requests.
To make it concrete, I use a simple framework that guides nearly every prompt I write: CREED.
The CREED Framework
Here’s what it stands for — and how it maps to real product work:
Context
Role
Expectations
Expand
Discover
These elements aren’t meant to be crammed into a single prompt. Working with AI should feel like a real conversation, not a one-time delegation.
Start by sharing context and discovering the topic. Ask questions. Refine your understanding. Then clarify your expectations, expand the response, and iterate. If the conversation starts to drift, don’t force it. Just start a new chat and reset the flow. Simple.
Context: Share insights and relevant analytics
Most people jump straight to the ask and get generic responses in return. Product managers know better: context is what sharpens every decision.
You sit at the center of customer feedback, usage data, and business goals. To get meaningful responses from AI, set the stage. Include the relevant signals, patterns, and product context that inform the problem.
Role: Ask the right expert
Knowing who to involve is a core part of product work. In your prompt, define the perspective you want the AI to take — whether it’s a UX researcher, marketer, analyst, or customer. Different roles see different problems. Prompting is no different.
Expectations: Be clear about what you need
A good prompt is like a good user story: it has a clear goal, a defined audience, and an expected output.
Spell out what you want and how it should be delivered: a short summary, a 5-bullet update, a comparison table. The sharper the ask, the sharper the response.
Expand: Explore, iterate, and refine
You rarely go with the first idea in a product workshop. Treat prompts the same way.
Request multiple versions, different framings, or constructive feedback. Use the results to iterate, combine, and converge on something sharper. Prompting is an active process, not a one-and-done instruction.
Discover: Ask leading, open-ended questions
Product managers excel at uncovering unknowns. Don’t jump to solutions.
Just like with any new initiative, start with discovery: ask thoughtful, open-ended questions to explore the problem space. Let the AI surface assumptions, challenge your framing, or suggest perspectives you might have missed.
An End-to-End Example
Let’s imagine you’re preparing for your first round of user research interviews for a new or underperforming feature. You’ve identified a specific problem area — for example, low engagement, high drop-off, or unclear user behavior — but you’re still in the discovery phase. You want to better understand user motivations, pain points, and context before jumping to solutions. Your goal is to craft a solid interview plan that helps you validate assumptions and uncover insights you might be missing.
1. Role + Context + Discover
I’m a product manager at a B2C EdTech startup working on onboarding. We’re seeing low attendance for trial lessons, and I’m preparing user interviews to explore why. What frameworks or techniques would you recommend for effective discovery?
This prompt sets a clear context (product area, target behavior, and problem), but doesn’t rush to a solution. Instead, it invites the AI to suggest discovery tools and frameworks—things like Jobs-to-be-Done, Bullseye Customer, The Mom Test, assumption mapping, or switching interviews. This mirrors how PMs approach the unknown: by opening the problem space first, not narrowing it prematurely. It positions the AI as a thought partner, not just a task executor.
2. Context + Role + Expectations
Act as a senior user researcher at a B2C EdTech startup. We’re seeing low attendance for trial lessons. Draft interview questions to uncover root causes. Use principles from The Mom Test and Jobs-to-be-Done to keep the questions open-ended, unbiased, and actionable.
This prompt gives the AI a specific role, a straightforward task, and relevant methodological guidance. By referencing The Mom Test and Jobs-to-be-Done, you’re steering the AI toward discovery-driven, non-leading questions that help uncover user motivations and unmet needs. You’re also providing just enough context to ensure the output aligns with your product environment. This combination of structure and flexibility allows the AI to deliver results that are both thoughtful and practically useful, just like working with a real researcher.
3. Expand
Review the interview questions above. What’s missing? How can they be improved or reworded? Suggest stronger alternatives where needed. Ensure the questions cover the following key aspects of our onboarding flow: X, Y, and Z.
This prompt presumes the AI acts as a collaborator, not just a generator. Treating the first draft as a starting point enables critique, improvement, and iteration, much like a product review. Asking the AI to identify gaps and suggest alternatives adds a second layer of quality control, making sure your research questions align with critical parts of the user journey. It shifts the AI’s role from writer to reviewer, enhancing both depth and usefulness.
Real Prompts I Use Every Day
Whether I’m refining a feature spec, summarizing a customer interview, or preparing a roadmap pitch, I rely on AI as a thinking partner. I use these real prompts in the middle of messy product work.
Refine an Epic, Feature, or User Story
Act as a senior product manager at a B2B e-commerce platform. Rewrite the feature proposal for clarity, structure, and completeness.
# Problem
* …# Evidence
* …# Solution Proposal
* …
# Customer Value
* …# Business Value
* …
Draft an Executive Memo
Based on the meeting notes below, write a concise summary for the executive team. Keep it to 5–7 bullet points, with a maximum of 7 words per point. Prioritize clarity, alignment, and key takeaways.
Summarize Customer Research
Act as a senior user researcher at a B2C EdTech startup. Based on the attached transcript and interview questions, extract the main insights. Summarize key user needs, frustrations, and patterns in clear, structured notes.
Market Research
List five competitors offering [X]. For each, summarize their key features, pricing model, and positioning in a single paragraph. Highlight any clear differentiators.
Critique
Act as a Chief Product Officer at a B2C EdTech startup, reviewing my Q3 roadmap proposal. Read the draft attached and critique it for clarity, prioritization logic, assumptions, and potential blind spots. Suggest improvements.”
How I Set Up My Prompting System
I’ve created a standard setup for most of my chats—partly to keep the tone consistent with my usual way of communicating, and partly to manage my own tendencies. I can be overly realistic (some might call it pessimistic) and tend to overshare. To stay focused and efficient, I use projects and custom instructions in ChatGPT or paste a setup at the beginning of a new chat.
Act as a Product Manager at [Company], focusing on [Product / Component].
Follow these writing style guidelines when creating content:
Formatting
* Use simple, clean formatting
* Include headers, bullet points, bold, and italic text as appropriate
* Avoid long, dense paragraphs—use white space to enhance readability
* Do not use emojis unless I explicitly say so
* Avoid overly complex or academic language unless the context demands it
* Skip decorative formatting or unnecessary symbols
Tone & Style
* Aim for clarity, brevity, and accessibility
* Use an engaging, professional, and confident tone
* Prefer active voice and direct language
* Optimistic and open-minded
* Free from complaints or pessimism
* Still radically candid and honest
Final Thoughts: Prompting Is Product Work
If you’re still treating prompting like a party trick, you’re leaving serious leverage on the table.
This isn’t about writing clever commands — it’s about thinking clearly, framing problems, and collaborating with a system that never sleeps. AI isn’t just a tool; it’s a teammate. One that can challenge your assumptions, tighten your writing, surface edge cases, and draft faster than you can finish your coffee.
The real shift? Stop prompting like a user. Start prompting like a product manager.
Because when you do, you’re not just typing into a box — you’re doing the work together.
