AI-powered work item generation for Azure DevOps describe your work in plain English and get a fully structured, ready-to-save item in seconds.
Built by Ansi ByteCode LLP
What It Does
Smart Work Item Creator adds a Create Work Item hub to your Azure Boards. Instead of filling out forms field by field, you write a plain-text description of the work you need to capture. Azure OpenAI turns that description into a complete, structured work item with a title, description, acceptance criteria (or repro steps for bugs), severity, and auto-generated child tasks or stories which you review, tweak, and save, all without leaving your board.
Supported Work Item Types
| Type | What the AI generates |
|---|---|
| Bug | Title, Description, Steps to Reproduce, Expected Behavior, Severity |
| User Story | Title, Description, Acceptance Criteria, child Tasks |
| Feature | Title, Description, Acceptance Criteria, child User Stories |
| Epic | Title, Description, Acceptance Criteria |
| Task | Title, Description, Acceptance Criteria |
Quick Start
1. Install the extension
Install from the Visual Studio Marketplace and grant the requested scopes (Work Items, Project, Graph). No Personal Access Token is needed; the extension uses your existing Azure DevOps browser session.
2. Configure Azure OpenAI
Click AI Settings (top-right of the hub) and enter:
| Field | Example |
|---|---|
| Endpoint | https://your-resource.openai.azure.com |
| API Key | Your Azure OpenAI key |
| Deployment Name | gpt-4o |
| API Version | 2024-02-01 |
Settings are saved in browser local storage no server, no shared secrets.
3. Create your first work item
- Open Boards → Create Work Item in your project's left navigation.
- Select the work item type from the tab bar.
- Paste or type a description of the work.
- Click Generate with AI.
- Review, edit, and click Save Work Item.
Creating a Bug
Select the Bug tab, describe the issue in your own words copy-paste a Slack message, an email, a stack trace, whatever you have and click Generate with AI.
The review modal opens with every field pre-filled:
- Title: concise, editable
- Description: full context, rich-text formatted
- Steps to Reproduce: numbered list inferred from your description
- Expected Behavior: what should have happened
- Severity: AI-suggested (1 Critical to 4 Low), adjustable
- Assignee: pick from your team
- Sprint / Iteration: select the target sprint
Auto-create Dev Task
When your project is configured to treat bugs as requirements (Project Settings → Boards → Working with bugs → Bugs are managed with requirements), a Bug Tasks section appears at the bottom of the modal. Toggle Auto-create Dev Task on and a child Task named Dev: <bug title> is automatically created and linked to the bug when you save.
Creating a User Story
Select the User Story tab, describe the feature or enhancement, and click Generate with AI.
The review modal fills in:
- Title and Description in standard "As a … I want … so that …" format
- Acceptance Criteria in detailed, bullet-pointed Given/When/Then statements
- Tasks in a numbered list of implementation tasks, each editable inline
- Assignee and Sprint / Iteration
Per-task Assignee
Expand any task card to edit its title, description, and assignee independently. Each task is created as a linked child work item when you save the story.
Key Features
Request Changes
Not happy with a field? Use the Request Changes box at the bottom of any review modal to ask the AI to refine specific fields — "make the acceptance criteria more concise" or "change severity to Critical" without regenerating the whole item.
Generate Child Items with AI
For User Stories, click Generate Tasks with AI inside the modal to produce a fresh set of child tasks from the story description. For Features, the same button generates child User Stories.
Parent Linking
Enter a Work Item ID in the Parent Work Item sidebar before saving. The extension resolves the item's title and type so you can confirm the right parent, then creates the hierarchy link automatically.
Attachments
Drag and drop files onto the Attachments drop zone (or click to browse). Files are uploaded to Azure DevOps and attached to the work item on save.
Rich Text Editing
All text fields in the review modal include a lightweight rich-text toolbar bold, italic, underline, ordered/unordered lists, and font size so you can format content before saving without switching to another tool.
No PAT Required
The extension calls Azure DevOps APIs using the authenticated browser session via the Azure DevOps Extension SDK. Your personal access token is never needed.
Settings Reference
| Setting | Where it lives | Notes |
|---|---|---|
| Azure OpenAI Endpoint | Browser local storage | Per-user, per-browser |
| API Key | Browser local storage | Never sent to any server other than your own Azure OpenAI resource |
| Deployment Name | Browser local storage | Must match the deployment in your Azure OpenAI resource |
| API Version | Browser local storage | Default: 2024-02-01 |
| Bugs as Requirements | Project Settings → Boards | Enables the Auto Dev Task toggle in bug modals |
Tips
- More detail → better output. The AI works best when you give it context: what broke, what you expected, which component is involved.
- Use Request Changes instead of regenerating. If 90% of the generated item is correct, use the change request box to fix the remaining 10% rather than starting over.
- Link a parent before saving. Setting the parent ID before clicking Save Work Item creates the hierarchy in one step — no need to edit the item afterwards.
- Attachments upload on save. Add screenshots or log files to the drop zone while reviewing; they will be uploaded and attached when you click Save Work Item.
Requirements
- Azure DevOps Server (on-premises) or Azure DevOps Services
- An Azure OpenAI resource with a GPT-4 class deployment (GPT-4o recommended)
Questions or feedback? Email us at [email protected]
Built by Ansi ByteCode LLP