AI agent for product owners
User Story Refinement Agent
Backlog items that are clear, testable and ready before the team picks them up
What it does
When stories are unclear, the team builds the wrong thing or stops mid-sprint to ask questions. Before each refinement session, this agent pulls the queued stories, linked designs and the team's definition of ready. It checks each story for a clear user and value, testable acceptance criteria, named dependencies and a size that fits one sprint. It drafts missing acceptance criteria from the story's intent and lists vague phrases as questions for the product owner. When a story is too large, it proposes a split into smaller stories that each deliver value. It then checks that every blocking dependency has an owner, and if one is missing it adds a question and checks again. It never marks a story ready on its own; the product owner approves each refined story. Edge case: a story that hides two different user needs is flagged for splitting by need, not just by size.
How it works
Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.
Read the steps as a list
- Stories queued for the next refinement session
- Pull the queued stories, linked designs and the team's definition of ready
- Check each story against the definition of ready
- Draft missing acceptance criteria and list ambiguous phrases
- Is each story clear, testable and sized for one sprint?If not: raise questions for the stakeholder or propose a split, then re-check. Back to step 3.
- Check dependencies on other teams and open questions
- Are all blocking dependencies named with an owner?If not: draft a dependency note and ask the owning team for a date. Back to step 6.
- Product owner approves the refined storiesThe agent waits here for your OK.
- Ready stories for sprint planning
How it decides
It measures each story against the definition of ready and proposes a split when a story is too large or mixes separate needs.
- Require testable acceptance criteria
- Split stories too big for one sprint
- Split stories that mix separate user needs
Make it yours
Every agent is a starting point. You choose these settings for your own situation.
- Definition of ready
- Acceptance criteria style
- Sprint size limit
- Split rules
What keeps you in control
It always asks you first
- Marking stories ready
Hard limits
- Does not mark stories ready without approval
- Does not invent requirements beyond intent
It stops when
- Done: stories meet the definition of ready
- Stop: the product goal is unclear
Set it up
We guide you through the set-up, step by step
Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.
- One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
- The agent then walks you through connecting your own data, one source at a time
- A downloadable copy with the flow chart, the rules and the full guide