Alibaba Cloud DevOps AI Assistant quickstart

Scenarios at a glance

Access AI Assistant

You can access AI Assistant from two entry points in Alibaba Cloud DevOps:

• Global entry: Click the AI Assistant icon at the top of the Alibaba Cloud DevOps page to start a conversation at any time.

• Contextual entries: Multiple pages provide contextual AI buttons (such as work items, code review, and pipeline editor pages). Click a button to start an AI Assistant conversation with pre-filled prompts for the current context.

Activate AI Assistant (administrator)

AI Assistant is disabled for all members by default. The organization administrator must activate the feature and configure access:

1. Log on to Alibaba Cloud DevOps and navigate to Admin Console.

2. Locate the AI Assistant settings and complete activation.

3. Specify which users can use AI Assistant. By default, no members have access. The administrator must manually authorize users (role-based authorization is not supported at this time).

Billing

Alibaba Cloud DevOps AI Assistant uses the core capabilities of StarOps (All-Domain Intelligent O&M Platform), so billing is handled by StarOps:

• Unit of measurement: Credits (AI_Credits).

• View usage: The organization administrator can view AI Assistant credit consumption on the Admin Console > AI Usage page.

• Bill source: Bills are issued by StarOps. For detailed billing rules, see StarOps billing information.

• Usage reference:

  • A code review (such as merge request AI review) costs approximately 100 credits per run.

  • A requirement optimization (such as smart requirement description optimization) costs approximately 50 credits per run.

Usage guidelines

• AI Assistant responds based on the current context (current project, current work item, current repository, and so on).

• For sensitive operations, AI only provides suggestions and does not execute automatically.

• When writing prompts, include: the goal, background, desired output format, and constraints.

Code review with AI

Entry point: Repository > Create merge request > ClickRequest AI Review

Example task: You submitted a merge request and want AI to perform an initial code review.

How to use:

This scenario does not require typing in the AI Assistant dialog box. After you click Request AI Review, AI Assistant posts review comments directly in the merge request. You can reply in the comments to ask AI to explain a specific finding or provide more detailed fix suggestions.

Expected result:

• AI lists potential bugs or risks in the merge request comments.

• AI provides optimization suggestions for code style, performance, and security.

• AI labels issue severity levels.

Next step: Fix issues based on the review comments, or @mention relevant colleagues in the comments for confirmation.

Generate pipelines with AI

Entry point: Pipeline list > ClickSmart Create Pipeline

Example task 1: You are new to Alibaba Cloud DevOps pipelines and unfamiliar with pipeline rules or YAML syntax. You want AI to generate a pipeline quickly.

Example task 2: Your team previously used Jenkins or GitLab CI and now wants to migrate to Alibaba Cloud DevOps. You want AI to convert an existing CI configuration into an Alibaba Cloud DevOps pipeline YAML.

Sample prompt:

Help me create a pipeline

Expected result:

• AI generates a runnable Alibaba Cloud DevOps pipeline YAML draft.

• AI explains the purpose of each stage and step.

• AI highlights parameters that need replacement (such as image address, environment variables, and repository URL).

Next step: Save and run the pipeline. If an error occurs, click Smart Troubleshoot — AI analyzes the cause, provides a fix, and can auto-fix the pipeline YAML.

Create scheduled AI tasks

Entry point: Alibaba Cloud DevOps Workbench > AI Assistant Tasks

Example task 1: Automatically send a daily iteration progress report for a specific project every morning.

Example task 2: Summarize weekly project work-hour data, compile a weekly report, and send it to the team.

Two ways to create a task:

Method 1: Create directly from the task list

In Alibaba Cloud DevOps Workbench > AI Assistant Tasks, click Create Task, set up the schedule, and describe the task in natural language. For example:

Every day at 9:00 AM, generate a progress report for project XXX from yesterday, and send it to DingTalk group XXX.

Method 2: Run the task in a conversation first, then convert to scheduled

1. In the AI Assistant dialog box, run a demo of the task first. For example:

   Generate a daily iteration progress report for project XXX from yesterday.

2. Refine the report content, format, and data scope through multiple conversation turns until the result is satisfactory.

3. Tell AI:

   Convert this report into a scheduled task that runs every day at 9:00 AM.

   Expected result:

• AI Assistant generates a report that meets your requirements.

• AI sets up the scheduled execution rule (such as daily or weekly, at a specific time).

• AI specifies the delivery target (such as IM group, email, or in-app notification).

Next step: Check the task execution history in the AI Assistant Tasks list to confirm the scheduled task runs as expected. Edit any time to adjust the content or schedule.

Import requirements with AI

Entry point: Project details page > Requirements list > ClickSmart Import Requirements

Example task 1: A product manager has a PRD document containing multiple features that need to be split into separate requirements in Alibaba Cloud DevOps, then assigned to the development team.

Example task 2: After a client meeting, you have a set of meeting notes containing multiple new requirements that need to be organized and imported into Alibaba Cloud DevOps for tracking.

Sample prompt:

@ProjectName Help me smart-import requirements

Then follow the AI Assistant instructions and provide the document. You can upload an attachment or provide an online document link.

Expected result:

• AI splits the document into multiple requirement items based on headings and structure.

• AI generates a title and description for each requirement.

Next step: Follow the AI prompts to select the requirement type, owner, iteration, and other fields, then start the import.

Optimize requirements with AI

Entry point: Requirement details page > ClickSmart Optimize Content

Example task: A requirement has only a brief description and needs content and structure optimization.

Sample prompt:

@RequirementID Smart-optimize the content

Expected result:

• AI identifies missing information in the requirement description and adds background and objectives.

• AI optimizes the content structure for clarity and readability.

• AI generates or improves acceptance criteria.

Next step: Review the optimized content and accept all or part of the changes. Continue asking AI to adjust as needed.

Generate test cases from requirements

Entry point: Requirement details page > Linked test cases list > ClickSmart Generate Test Cases

Example task: You have a requirement work item and need to quickly generate test cases covering positive, negative, and boundary scenarios.

Sample prompt:

@RequirementID Smart-generate test cases

Expected result:

• AI generates multiple test case titles.

• AI assigns priority levels (P0/P1/P2).

• AI provides preconditions, steps, and expected results.

Next step: Import all test cases with one click, or selectively keep specific ones.