GitLab has introduced agent orchestration within its Duo Agent Platform to keep review, tests, security scans, and approvals moving without manual handoffs between development stages. The company notes that while AI coding tools speed up code creation, delivery can still stall at verification, review, and ship steps, which are often labor-intensive and time-consuming.
HOW IT WORKS: END-TO-END GOAL-DRIVEN FLOWS
Goal-driven flows let teams specify an objective and allow the flow to run end-to-end—from review and tests to security checks and approvals—using the Duo CLI, headless mode, or Agentic Chat. A separate model verifies progress at each step, and teams can stop or redirect a run if needed. TechStaged has also covered GitLab arms developers with cost controls for AI agents via Duo Agent Platform and Orbit.
- Start a goal-driven flow with /goal in Duo CLI or Agentic Chat
- Flow runs under the same identity, policy, and evidence chain as other GitLab activity
- Flows can be started from Slack conversations or from GitLab events such as mentions, assignments, or pipelines
KEY CAPABILITIES FOR FASTER DELIVERY
GitLab highlights several capabilities that reduce wait times and maintain governance across the pipeline. These include: automated handoffs across review, tests, scans, and approvals; Slack-based initiation that starts from a discussion thread; and Custom Flows that reflect a team’s existing processes.
- GitLab for Slack starts work from conversations
- Custom Flows define multi-step sequences
- Flows run under the same identity, policy, and evidence chain
AVAILABILITY AND PREREQUISITES
The features are available on GitLab.com, GitLab Self-Managed, and GitLab Dedicated, on Premium and Ultimate tiers. They require GitLab 19.3 or later and Duo CLI 9.17.0 or later.
CONTEXT AND NEXT STEPS
Foundational agents and flows cover plan through analyze stages in the GitLab UI, IDE, and Duo CLI, with further integrations available via the AI Catalog, MCP client/server, and GitLab for Slack. The announcement references a broader rollout announced at GitLab Transcend and positions agent orchestration as the backbone of a more automated software factory within GitLab’s ecosystem.
RELATED COVERAGE
- GitLab arms developers with cost controls for AI agents via Duo Agent Platform and Orbit
- GitLab Credits Bring Usage-Based Pricing to Duo Agent Platform
- GPT-6 Astra on GitLab Duo Agent Platform delivers faster runs and fewer tokens
- GitLab scales Git for the agentic software factory with next-gen SCM
- Developer Tools articles
SOURCES
- GitLab: GitLab's agent orchestration eliminates the wait between development stages Published · Primary source
