How to use AI for User Stories and Acceptance Criteria?
BLOG 3: Is Your User Story Really Ready? – use AI as a Second Pair of Eyes
A User Story can look perfect. “As a user, I want…, so that…” Everything is in place.
Then the development team asks one question: “And what should happen in this case?” And you realize the story isn’t quite as clear after all.
That is normal. Product Management is not about writing perfect tickets. The real challenge is building a shared understanding of the problem, the user, and the expected outcome.
AI can read a User Story and ask:
- Is the user clearly defined?
- Is the problem clear enough?
- Is the expected outcome concrete?
- Which business rules are missing?
- Which assumptions are not explicitly stated?
- Which scenarios might be important?
This is far more useful than simply saying: “Write me Acceptance Criteria.”
Proposed prompt to get started
Role: You are an experienced Product Manager / Product Discovery partner. Help me check whether the User Story and the expected outcome are defined clearly enough. Do not make final decisions on behalf of the Product Manager or the team.
Context: [describe]
Feature: [name]
User: [who uses it]
Problem/Need: [describe]
Expected outcome: [describe]
Business rules/Constraints: [enter]
Scope/Out of scope: [enter]Based on this information:
- Suggest or improve the User Story.
- Suggest relevant Acceptance Criteria.
- Do not invent business rules.
- If something is not defined, list it as a question.
- Clearly distinguish between:
- Acceptance Criteria;
- test scenario;
- technical requirement;
- question for clarification.
Use Given–When–Then only when it adds clarity.
At the end, answer:
- Does the Story clearly describe the user, the need, and the outcome?
- Do the Acceptance Criteria clearly define what it means for the Story to be acceptable?
- Which important decisions are still undefined?
ATTENTION: The most important part of AI prompt is not the acceptance criteria.
This might be the most important lesson.
If AI generates 15 Acceptance Criteria, that doesn’t mean the User Story is better. It might just be longer.
The real value lies in the question: “What is still undefined?”
For example, AI can show that you actually don’t know:
• what the user sees when there is no data;
• who has permission to perform the action;
• what happens when a business rule is not met;
• whether a specific behavior is even in scope.
These are product questions. And it is much better to raise them during refinement than to discover them mid-development.
ANOTHER ATTENTION: Acceptance Criteria ≠ Everything That Can Be Tested
This is another common mistake. Not every possible scenario is automatically an Acceptance Criterion. Some scenarios are:
• questions;
• test cases;
• technical requirements;
• decisions that still need to be made.
A good AI prompt should make that distinction. Otherwise, you will end up with a massive list of “criteria” that only gives a false sense of completeness.
Try it today
Take one User Story that often gets bounced back to the team. Run the prompt.
But first, look at only one part of the output: “What is unclear or missing?” If AI finds something that neither you nor the team explicitly defined – you already have value.
Once you know what needs to be built, one of the toughest product questions follows: What should be built first?
Here, AI can help make prioritization more transparent – but it cannot decide for you.
BLOG 2: How to use AI for Epic Decomposition
In the previous part, Don’t Start from a Blank Page: Use AI to Find Vertical Slices, we showed how do you turn a big product problem into smaller pieces your team can understand and deliver. This is where many teams end up staring at a blank board, and AI can be a useful thinking partner.
BLOG 4: How to Use AI for RICE and Checking Assumptions
Clarifying Acceptance Criteria ensures you build the feature right. Next, you need to make sure you are building the right feature first. And find out how AI can help.