/
Cascading v2: Cloud AI implements Figma and Jira

Cascading v2: Cloud AI implements Figma and Jira

  • Verified

    • Learn how to use a cloud coding agent (Github Copilot) to implement a Jira story with Figma links and see the result in a Github Codespace.

    We are also able to do much of this with Gitlab using Bitovi’s AI-Coding-Agent (claude-based) and MR environments. Stay tuned for that setup!

    Did you know … you can hire Bitovi to set up your Cascading infrastructure and train your team on how to use it.

    Setup:

    Contents:

    https://www.youtube.com/watch?v=7mucYeLCvVc

     

    👉 Bitovi can help you integrate this into your own SDLC workflow: AI for Software Teams

    The Workflow

    1. A product owner and designer work together to create:

      1. Figma designs

      2. A Definition of Ready story that links to Figma designs within relevant acceptance criteria.

    2. The product owner

      1. assigns the story to an AI Agent user (e.g. BitoviAI), and

      2. puts the ticket In Progress

    3. The AI Agent creates a pull request and implements the feature

    4. The story’s status is changed to Review

    5. The product owner and designer are able to see the live app.

     

    1. Creating a Story that an AI can implement

    There’s a lot of ways to write a story. See Story Writing and Definition of Ready for Bitovi’s best practices.

    For this AI workflow, it’s critical to write detailed acceptance criteria that link to the Figma designs that show how the application should look in that state. We suggest using a nested pattern:

    Acceptance Criteria

    Getting Figma Links

    1. Click on the frame you want to reference. You’ll see a blue border show up.

    2. Right-click and select “copy as” > “copy link to selection”

    3. Paste that link in Jira.

     

    2. Kicking off the AI Agent Implementation in Jira

    To tell the AI agent to start coding the Jira ticket:

    1. Set the assignee to [Company]AI

      image-20250912-192333.png

    2. Set the status to In Progress

      image-20250912-192419.png

    In a moment, a comment with a link to the github issue will be made.

     

    3. Monitoring the development progress

    Jira submits an issue that the AI coding agent will use as a reference to build the code. A link to that issue will be added to the Jira issue’s comments:

    You can click that and see the GitHub issue:

    In the GitHub issue, you can find a link to the pull request where the AI agent is making the actual code. Click the link highlighted above to see the pull request below.

    Next, you can click View Session as shown above to see the progress of the AI coding agent’s work:

    4. Knowing when the Agent has completed the implementation

    Besides monitoring the session above, when the agent’s work has completed, the status of the ticket will be changed to Review. It’s possible to change the automation to also assign a particular person the ticket too.

     

    5. Testing the implementation

    To run the app directly from GitHub’s website, use a Codespace which will launch from Copilot’s commit.

    From the PR page, navigate to the Commits tab:

    There should be 2 commits, the first is simply a placeholder, the second is what we’ll be opening up in a Codespace. Click the second commit:

    From here, select Review in Codespace:

    This will open up a Codespace with the changes Copilot made in the commit. The app should build and run automatically: