Story Agent
The Story Agent turns a need expressed in plain language -- "Today the duplicate rule only checks the account name, it should also check the company registration number" -- into ready-to-refine user stories. Each story has a role, a need, a goal and Given / When / Then acceptance criteria, and is based on what your latest aprity scan found in the org.
Where to use it
| Surface | What you do there |
|---|---|
| aprity web portal | User stories in the AI assistants group of the portal menu. Generate, refine and validate stories, then export them as Markdown. |
| AI clients (MCP) | The generate_user_story tool, when a user explicitly asks an AI client for a user story. See Remote MCP Server. |
All surfaces use the same Story Agent and the same scan data.
How it works
- Choose a type of change (portal): Extend, Modify, Remove, Rename or Optimize.
- Describe the need in plain language, as you would to a colleague.
- Answer the agent's questions. When the request is ambiguous, the agent asks short clarifying questions before writing anything.
- Review the impact map. The agent shows what already exists in the org and which fields, rules, screens, processes, automations and permissions the change touches.
- Review the stories. For a large change, the agent proposes a breakdown into an epic, features and stories. Each story comes with acceptance criteria written in Given / When / Then form.
- Refine, validate, export. Ask for changes in the conversation ("split into sub-stories", "add a criterion for inactive accounts"), then validate. In the portal, export the validated stories as Markdown.
References that the agent could not find in the scan are flagged as unverified, so a reviewer can spot them before the stories reach a sprint.
The agent does not read your org live. It only sees the latest successful scan. Run a new scan first if a recent change is missing.
What a story contains
- As a / I want / So that -- the role, the need and the expected benefit.
- Acceptance criteria in Given / When / Then form, based on the automations and validation rules the scan found.
- Impact -- the Salesforce components the story touches.
For larger changes, stories are grouped under an epic and features, so the backlog can be sequenced without slicing the work horizontally (all fields first, all automations later).
Credits
The Story Agent consumes credits from your organization's budget:
| Action | Credits |
|---|---|
| Conversation message | 1 |
| Generating stories | 3 |
| Regenerating stories | 2 |
| Plan | Credit budget |
|---|---|
| Documentation | Not included |
| Intelligence | 100 per month |
| Trial | 6 for the whole trial |
The portal shows the credits remaining this month. On Intelligence, credits reset on the first day of each calendar month and unused credits do not roll over. When the budget is exhausted, new requests are refused until the next reset.
See Plan Comparison for the full feature matrix.
Limitations
- Stories reflect the latest successful scan, not the live org.
- Generated stories are a first draft. Have a business analyst or product owner review them before committing them to a sprint.
Troubleshooting
| Issue | Resolution |
|---|---|
| "The Story Agent is not available" | Your plan does not include the Story Agent. See Plan Comparison. |
| Stories feel outdated | Run a new scan, then start a new discussion. |
| Credits exhausted | Wait for the monthly reset, or contact contact@aprity.ai. |
| Stories miss objects you asked about | Confirm the scan covered those objects. Widen the scope in Analysis Configuration if needed. |
For anything else, open a ticket from Support in the portal account menu, or contact contact@aprity.ai.