GitLab has launched the public beta of its GitLab Duo Agent Platform, an orchestration tool that enables developers to collaborate asynchronously with AI agents across the DevSecOps lifecycle. The platform, now available to GitLab.com Premium and Ultimate customers as well as self-managed installations, transforms traditional, linear development workflows into dynamic, multi-agent systems where AI handles routine tasks such as refactoring, security scanning, and research, while developers focus on complex problem-solving.
At its core, the Duo Agent Platform leverages GitLab’s position as the centralized system of record, encompassing code, merge requests, CI/CD pipelines, test results, compliance checks, and project history, to provide AI agents with full organizational context. This allows agents to make informed, autonomous contributions aligned with development standards and workflows.
The public beta rollout introduces several key features designed to enhance collaboration between developers and agents. Agents can now operate in parallel, with different roles such as a Software Developer Agent, a Security Analyst Agent, or a Deep Research Agent, each performing specialized tasks under unified orchestration. Developers can interact with these agents through chat interfaces embedded directly in VS Code, JetBrains IDEs, and GitLab’s Web UI, delegating tasks and receiving feedback in natural language with commands such as “/explain”, “/tests”, or “/include”.
The platform also allows organizations to define custom agent rules using natural language, ensuring that agent behavior, coding preferences, and compliance requirements align with established standards. Beyond GitLab’s own ecosystem, the Duo Agent Platform integrates with the Model Context Protocol (MCP), enabling agents to interoperate with external systems and services. This positions the platform as not only a developer productivity tool but as an extensible, intelligent layer across the entire DevSecOps ecosystem.
GitLab plans to update the platform monthly as part of its 18.x release train, with general availability expected later this year. Early industry feedback has been positive. Bal Kang, Engineering Platform Lead at NatWest, stated that Duo agents embedded in their software development lifecycle have already “boosted productivity, velocity, and efficiency” by freeing developers from repetitive tasks. Rachel Stephens, Research Director at RedMonk, added that GitLab’s orchestration platform represents “a significant step forward in integrating AI agents into existing DevSecOps toolchains”.
By introducing orchestration, developer context, and intelligent automation into a unified, agent-based framework, GitLab’s Duo Agent Platform public beta offers a glimpse into the future of software engineering collaboration.
But GitLab is not alone. Rivals including GitHub, JetBrains, and Atlassian have already deployed big agent-driven solutions, each taking a distinct approach to integrating AI into the development lifecycle.
GitHub’s Copilot, powered by OpenAI Codex and later GPT-4 and now GPT-5, popularized the idea of AI as a real-time “pair programmer”. Unlike GitLab’s orchestrated, multi-agent model, Copilot focuses on inline code generation and developer assistance within the IDE. GitHub has also introduced Copilot Workspace, an experimental platform where AI agents handle entire workflows, such as project setup, dependency updates, and CI configuration (GitHub Blog). While GitLab emphasizes orchestration across multiple agents with organizational context, GitHub remains closer to the developer’s keyboard, prioritizing productivity and rapid coding suggestions.
JetBrains took a different route with the AI Assistant for IntelliJ-based IDEs. Rather than building a multi-agent orchestration layer, JetBrains integrated AI features, such as smart code explanations, inline documentation, and automated refactoring, directly into the IDE’s workflow. This approach resembles GitHub’s, but keeps everything IDE-centric. JetBrains positions AI less as a “collaborative agent ecosystem” and more as an extension of the development environment, reinforcing its philosophy that developer productivity is best improved by tightening the loop between coding and feedback inside the IDE.
Atlassian has leaned on AI agents to streamline project management and collaboration. Through Atlassian Intelligence, Confluence and Jira now use AI to summarize issues, generate documentation, and automate routine administrative tasks. Unlike GitLab or GitHub, Atlassian’s AI agents don’t write code; instead, they focus on removing the overhead of planning, documentation, and ticket management. The strategy is less about developer-agent pair programming and more about ensuring teams can move work through the pipeline with fewer bottlenecks.
While GitLab differentiates itself with its agent orchestration platform, compared to these competitors, all companies are converging on the same vision: AI agents as core participants in the software development lifecycle.
