The Unsexy Way to Deploy an AI Agent

Description and use

Automating AI workflows with GitLab CI/CD: zero rewrite, repository and Git history access, secrets, and scheduling.

Problem

Deploying AI workflows that rely on local repository context to the cloud without manual triggering or excessive infrastructure changes.

Outcome

Zero rewrite of the existing workflow; the CI/CD pipeline automatically provides repository context, Git history, secrets, and scheduling, enabling fully automated cloud execution.

Workflow

• 1. Store the AI workflow in a GitLab repository.
• 2. Configure a scheduled CI pipeline to trigger the workflow at desired intervals.
• 3. The CI runner checks out the repository and accesses Git history automatically.
• 4. The pipeline runs the workflow scripts, using available CI secrets and dependencies.
• 5. The workflow produces an investigation report and exits.
• 6. Repeat on schedule without manual intervention.

Evidence

I built an internal investigation agent that periodically picks up newly triaged engineering issues, investigates the relevant codebase and Git history, and produces an investigation report.

Source

Instead, I put the existing workflow behind a scheduled CI pipeline.

Source

Zero rewrite.

Source

The CI runner already has most of this. It can: check out the repository, access Git history, run scripts, access CI secrets, install dependencies, run containers, call external APIs.

Source