Integrate Applications and Workflows with CI/CD

A continuous integration and continuous deployment (CI/CD) process maintains supported application and workflow source artifacts in Git. Changes are reviewed, packaged, deployed to a target environment, and tested. Version control provides a history of changes and supports collaborative development.

Key Capabilities

The CI/CD process includes the following activities:

• Version-controlled source: Maintain supported application, workflow, and related metadata files in the project Git repository.
• Reviewed changes: Use branches and merge requests to review changes according to your project’s approval requirements.
• Packaging and deployment: Track package creation and deployment to each target environment separately.
• Deployment verification: Confirm that the expected artifacts and intended version are available in the target environment, then validate their behavior.

Prerequisites

Before you begin, confirm that you have access to the project repository and review process, the supported deployment tooling, and the target AI Agent Studio environment. Make sure the environment and tooling meet the requirements for your release and that the account used for testing can execute the relevant applications and workflows.

Follow your project’s repository and approval conventions. Review pipeline triggers and target environments before running a deployment.

Scope

Include only artifact types supported by your release and deployment method. Review references between applications, workflows, and related artifacts, and ensure that required dependencies are available in the target environment.

Use CI/CD in AI Agent Studio

Here's how you can use CI/CD in AI Agent Studio:

1. Prepare the source change.

   Create a branch for the change and update the source files using the required repository structure. Validate artifact syntax and required properties, review dependency references, and run the applicable tests before submitting the change for review. Commit the source changes and associated tests.

2. Submit and review the change.

   Create a merge request for the target branch. Describe the application or workflow change, include validation results, and obtain the approvals required by your project before releasing the change.

3. Build and deploy the approved change.

   Run the configured pipeline for the approved repository revision. Review build and deployment results for each target environment, including failed or skipped artifacts. Confirm that the expected artifacts are available and that the intended version is active. If a failure occurs, retain the deployment reference and relevant logs, correct the issue, and rerun the affected process.

4. Validate the deployed application.

   Run representative application and workflow scenarios in the target environment. Verify required dependencies, permissions, live integrations, expected responses, and business actions. When automated tests replay recorded or synthetic data, perform separate checks of the current live integrations and data.

Best Practices

• Use short-lived branches, and describe merge requests in terms of their outcomes.
• Record the source revision, deployment reference, and test-report links with the change.
• Verify expected artifacts and runtime behavior, even when the pipeline reports success.
• If build or deployment fails, preserve the build number and console-output reference, correct the source or configuration, and run a new build.
• Review supported artifact types and dependencies before deployment.