Part 2... Can AI Revolutionize User Story Creation?

Dear subscribers, continuing from last week article; AI agents can create just basic outline,

at least for now.

Let's further explore?

AI in Context of the 3Cs Approach

When working with the 3Cs, here’s how AI tools like Copilot or ChatGPT can play a supporting role:

Card (Outline and Placeholder):

• AI is quite effective in drafting the initial outline of a user story. For example, generating the "As a [user], I want [functionality] so that [value]" format is straightforward for AI.
• However, the AI-generated story will only serve as a placeholder or starting point, requiring the team to bring in context and depth during refinement sessions.

Conversation (Backlog Refinement and Discussions):

• This is where the real value lies, and AI is not designed to replace the collaborative nature of these discussions. Your practice of having thorough conversations with the PO ensures that the story is enriched with context, clarifies the “why” behind the request, and sets realistic expectations.
• AI might help prompt questions or surface considerations (e.g., dependencies or potential risks), but it lacks the ability to deeply engage in the human-centered discussion that drives clarity.

Confirmation (Acceptance Criteria and Definition of Done):

• While AI can suggest generic acceptance criteria, these tend to be surface-level and may not capture the nuances of the system or feature being developed.
• As you’ve mentioned, deriving acceptance criteria collaboratively during refinement is critical. This ensures that all edge cases, business rules, and scenarios are well-documented and understood.

My Experience with ChatGpt

The fact that ChatGpt provided only a basic outline aligns with many teams’ experiences. AI tools can handle boilerplate tasks but don’t yet have the capability to match the depth of a good human conversation or refinement session. This makes them more of a productivity aid than a strategic contributor.

Suggestions for Leveraging AI in Your Workflow:

Given your already-strong practices around backlog refinement and collaboration, here are some ways to make AI more useful:

• Pre-Refinement Preparation:
• Use AI to generate draft user stories for new features or epics. These drafts can act as conversation starters during refinement.
• Have AI generate questions or considerations based on the user story, helping the team explore scenarios or identify potential gaps.
• Acceptance Criteria Templates:
• AI can generate initial acceptance criteria templates that you can then review, refine, and align with your specific context.
• For example, ask the AI to focus on certain non-functional requirements, such as performance or accessibility, if relevant to your story.
• Spike Stories:
• For exploratory work, use AI to help structure “spike” user stories that outline objectives and research questions.
• Collaborative Ideation:
• Use AI in brainstorming sessions to generate multiple perspectives or ideas for features, even if they aren’t directly actionable. It can help the team think outside the box.
• Documenting Conversation Outcomes:
• After backlog refinement, AI can help quickly document the final version of the user story, including acceptance criteria, based on inputs from the team.

Final Thoughts:

Your strong practice of engaging in meaningful backlog refinement and discussions with the PO is what ensures high-quality user stories. AI tools like ChatGpt or Copilot are better suited to handle the mechanical parts of story creation, but they cannot replace the human collaboration and shared understanding that emerge during conversations.

AI should complement, not substitute, the process—providing efficiencies for repetitive tasks but leaving the strategic and contextual work to your team.

Would you consider using AI to generate story prompts or draft acceptance criteria to save time in the early stages of refinement? Or do you find its limitations outweigh the potential benefits in your context?