Use Ducalis’s built-in AI chat or your own assistant through cloud MCP to draft requirements, user stories and testable acceptance criteria from the requests you have collected. Your team reviews the draft and connects it to Jira or Linear work.
Draft requirements with AIUse the source Idea, supplied customer context and existing work to explain the problem. Keep the evidence connected to the requirements.
Review scope, user stories, acceptance criteria and unanswered questions. Confirm the issue and destination before creating or updating work.
Select the source feedback and ask for a structured draft. Choose the chat you work in; keep the same team review before using the result.
Open the AI Assistant beside your boards. Ask questions about feedback, draft requirements or configure a workflow, then review the result with your team. You can start here without connecting an external MCP client.
See the built-in AI Assistant →Work with Ducalis from a compatible AI client such as Codex or Claude Code. Use it from your desktop, or your phone when the client supports remote MCP. Add the server URL from your API, MCP settings and authorize the connection in your browser through OAuth.
See MCP connection steps →An Acme customer asks to export invoices as CSV for their accountant. Your team selects the source request and asks the assistant to draft a problem statement, scope, user story and acceptance checks.
┌─ LINK REQUIREMENTS TO WORK ──────┐ │ source Idea CSV-17 │ │ issue Linear APP-218 │ ├──────────────────────────────────┤ │ Link existing or create new │ │ Keep reviewed scope and checks │ │ Verify the chosen issue │ │ Source request stays linked │ └──────────────────────────────────┘
Ask your assistant to work from a specific request or issue. Include the customer problem, relevant feedback links, known constraints and the work that already exists.
The assistant can use Ideas, issues, comments and other accessible Ducalis account data. It operates within your existing permissions. Give it the scope for this draft and ask it to flag missing information.
Use a consistent issue-description structure: problem, audience, scope, non-goals, user stories, acceptance criteria, source links and open questions. The result can be a structured PRD draft or a reviewed issue description, depending on the work you need.
For the sample CSV request, accounting-system sync is a proposed exclusion. The exact CSV columns remain an open question until the team confirms them.
Organize the source feedback → · Find recurring customer problems → · Review duplicates and scope first →
┌─ TESTABLE ACCEPTANCE CHECKS ─────┐ │ Given authorized billing user │ │ and a selected date range, │ │ when they export invoices, │ │ then CSV matches that range. │ ├──────────────────────────────────┤ │ Restricted user: no export │ │ No rows: clear empty state │ │ Suggested checks · review │ └──────────────────────────────────┘
The source asks for invoice CSV export. The following structure is an illustrative proposal that the team reviews and completes.
| Part of the draft | Illustrative content |
|---|---|
| Supplied problem | The customer needs to share invoice information with their accountant as a CSV file. |
| Proposed requirements | Provide an export action in Billing; export the selected invoice date range; produce the agreed CSV columns. |
| User story | As a billing admin, I want to export invoices for a selected date range so I can share them with my accountant. |
| Acceptance criterion | Given an authorized billing user and a selected date range, when the user exports invoices, then the CSV contains the invoices for that range with the agreed columns. |
| Permission and empty state | A user without export permission cannot export. If the selected range has no invoices, show the agreed empty-state behavior. |
| Proposed non-goal | Automatic sync with an external accounting system. |
| Open questions | Which columns and date format are required? Who can export? How should failed or large exports behave? |
Check every factual statement against the source. Product reviews the problem, audience and scope; engineering checks feasibility and technical questions; QA reviews whether the acceptance criteria can be tested.
Edit the suggested behavior, remove unsupported assumptions and resolve or retain open questions. Keep unknown dependencies and effort estimates visible until the team evaluates them.
For an existing issue, Rewrite Issue proposes a new description. Review it before Update Issue; Try again requests another version and Cancel keeps the original. When generating new work from an Idea, review before Create Issue.
Keep the approved text and remaining questions with the task. An AI draft is an input to the team’s decision; it does not establish that the feature is ready for implementation.
┌─ TEAM CHECKS THE DRAFT ──────────┐ │ Check facts against sources │ │ Confirm scope and exclusions │ │ Review permissions and checks │ │ Resolve or keep open questions │ ├──────────────────────────────────┤ │ Approve the text to use │ │ Creation destination confirmed │ └──────────────────────────────────┘
If matching work exists, link the customer Idea to that issue. If it is new work, use + New to create an issue in Ducalis or your connected tracker, with the reviewed description.
In the example, the request is connected to Jira BILL-142 or Linear APP-218. Confirm the right issue and destination, check the resulting description in your chosen workflow and keep the original Idea link available.
New-issue creation, existing-issue edits and tracker Sync Back are different operations. Use your configured integration for the work you are updating and verify the fields that reach the destination.
Jira backlog workflow → · Linear backlog workflow → · Configure your task tracker →
┌─ LINK REQUIREMENTS TO WORK ──────┐ │ source Idea CSV-17 │ │ issue Jira BILL-142 │ ├──────────────────────────────────┤ │ Link existing or create new │ │ Keep reviewed scope and checks │ │ Verify the chosen issue │ │ Source request stays linked │ └──────────────────────────────────┘
Clearer scope gives the team better input for estimates, priorities and planning.
Compare the reviewed work with other customer requests, bugs and improvements. Keep evidence, uncertainty and criterion definitions visible.
Prioritize the backlog →Use team evaluations and Alignment to find differences, clarify the requirements and revisit estimates when needed.
Align the team on priorities →When the team chooses what to build, set dates and dependencies in Gantt. Review the release window and customer-facing updates separately.
Prepare the release plan →Use these references to review the problem, document structure and connection to tracker work.
Choose the source feedback, ask for a structured draft and review its scope, story and acceptance criteria. Connect the reviewed result to the right issue.
Draft requirements with AIOpen the AI Assistant guide → · Choose built-in chat or cloud MCP →