Session 5 — Azure DevOps Integration
Phase: 2 — Enablement & Integration Target Week: Week 4 Duration: 2.5 hours Format: Hands-on workshop with pipeline-level changes Audience: Turner development team, DevOps/platform engineers (required)
Objective
Embed AI workflows into the learning bed's Azure DevOps pipelines so the lift carries from individual developer practice into the team's shared pipelines. By the end of the session, the learning bed has at least one AI-integrated pipeline step running — and the pattern is documented for replication across the portfolio.
Key Topics
- Backlog management: using AI to refine user stories, decompose epics, and draft acceptance criteria
- Pull request workflows: AI-assisted PR description authoring, change summaries, and reviewer prompts
- Automated code vetting in pipelines: lint/security/style checks augmented with AI judgment
- Test generation and coverage gap analysis as a pipeline step
- Safe patterns for AI in CI/CD: scoped permissions, deterministic gates, human-in-the-loop checkpoints
- Failure modes to avoid: flaky AI checks, non-deterministic gates, prompt leakage in logs
- Cost and rate-limit considerations when scaling AI into pipelines
Outcomes
- At least one AI-augmented pipeline step running in the learning bed's pipeline (PR summarization, automated review, or test scaffolding)
- A documented pattern Turner can replicate across additional pipelines
- Clear list of pipeline-level guardrails reflecting the Session 2 governance framework
- Integration patterns added to the Developer Reference Guide
Inputs Required from Turner
- Permission to add or modify the learning bed's pipeline (non-production branch is fine)
- DevOps engineer authorized to commit pipeline changes
- Service account or token for AI API access from pipeline agents
Deliverable Contribution
Produces the substantive content for the Guidance on Integrating AI Workflows into Turner's Development Environment and DevOps Pipelines deliv