GitLab rolls out Duo AI Agent Platform for DevOps teams

GitLab has launched GitLab Duo Agent Platform for general availability, positioning the product as a way for software teams to run AI agents across more of the software delivery process.

The company said the product is available to GitLab Premium and Ultimate customers on GitLab.com and on Self-Managed deployments. GitLab said it expects availability on GitLab Dedicated during the GitLab 18.8 release cycle.

GitLab framed the release around a gap between faster code writing and overall delivery speed. The company said developers spend a minority of time writing code. It said faster coding can shift pressure to areas such as code review queues, security checks, compliance work, and bug fixing.

"The general availability of GitLab Duo Agent Platform marks a fundamental shift in how AI delivers value in software development," said Manav Khurana, Chief Product And Marketing Officer, GitLab.

"We've seen AI make coding faster, but that is just one part of what it takes to deliver innovation at scale. Now organisations can orchestrate AI agents across the entire lifecycle in one unified system, with the comprehensive context and governance they need to help them innovate and ship software faster," said Khurana.

Product scope

GitLab Duo Agent Platform combines several elements that GitLab groups under an "agentic" AI umbrella. The company said the platform includes agentic chat, a set of prebuilt agents, workflow automation, and administrative controls.

GitLab said Agentic Chat works across the GitLab web interface and integrated development environments. GitLab said the tool draws on information from items such as issues, merge requests, pipelines, and security findings. It said Agentic Chat uses multi-step reasoning for complex questions and can take autonomous actions.

GitLab listed use cases across analysis, coding, CI/CD, and security work. It said the chat tool can create issues and merge requests and summarise project context in the GitLab web interface. It said the tool can assist with understanding unfamiliar code and dependencies within an IDE or repository.

GitLab said Agentic Chat can generate code, configurations, and Infrastructure-as-Code across multiple languages and frameworks. It said the tool can fix bugs, modernise code, generate tests, and produce documentation. GitLab said the tool works with VS Code, JetBrains IDEs, Cursor, and Windsurf.

In CI/CD, GitLab said the chat tool can explain, configure, and troubleshoot pipelines. For security work, GitLab said it can explain vulnerabilities, prioritise issues based on reachability, and recommend fixes.

Agents and flows

The general availability release also includes what GitLab calls Foundational Agents. GitLab said it built these agents for complex tasks that sit outside direct code authoring. It listed two agents at general availability.

GitLab said the Planner Agent structures and prioritises work within GitLab. GitLab said the Security Analyst Agent reviews vulnerabilities and other security signals and explains impact in plain language.

The platform also supports Custom Agents through an AI catalog that GitLab described as a central repository. GitLab said teams can create, publish, manage, and share custom agents and flows across an organisation. GitLab said custom agents can use specific context and follow internal engineering standards and guardrails.

GitLab also highlighted External Agents that integrate into GitLab. It cited Claude Code from Anthropic and Codex CLI from OpenAI. GitLab said users can access these tools within GitLab for tasks such as code generation, code review, and analysis.

The company also described Foundational Agentic Flows that chain multiple agents. It listed flows that start from an issue and produce a merge request, migrate or modernise pipeline configuration, analyse CI/CD failures, analyse and streamline code review activity, and guide development work within IDE workflows.

Controls and visibility

GitLab placed governance controls alongside the agent tooling. It said the platform provides visibility into how agents are used and what actions they perform. GitLab said usage and activity details can show adoption and measure impact.

GitLab said top-level namespace owners can select a model, with inherited settings for subgroups. It said GitLab Self-Managed deployments can use self-hosted models. GitLab also said administrators can set group-based access controls at the namespace level.

GitLab said the platform integrates with LDAP and SAML. It presented these integrations as part of governance and scaled administration.

Pricing model

The company also introduced a usage-based charging structure for the new products. GitLab said it will use GitLab Credits, described as a virtual currency. GitLab said Premium customers will receive USD $12 of included credits per user each month, and Ultimate customers will receive USD $24 per user each month.

GitLab said customers consume credits when they use GitLab Duo Agent Platform features. It said customers can purchase additional credits either through a shared pool or on-demand monthly spending.

A NatWest engineering leader described the product's impact in terms of workflow integration.

"GitLab Duo Agent Platform enhances our development workflow with AI that truly understands our codebase and our organisation," said Bal Kang, Engineering Platform Lead, NatWest.

"Having GitLab Duo AI agents embedded in our system of record for code, tests, CI/CD, and the entire software development lifecycle boosts productivity, velocity, and efficiency. The agents have become true collaborators to our teams, and their ability to understand intent, break down problems, and take action frees our developers to tackle the exciting, innovative work they love," said Kang.

Industry analyst firm IDC pointed to a wider shift towards agent use in DevOps pipelines.

"IDC forecasts that by 2030, 70% of organisations will embed AI agents into DevOps and DevSecOps pipelines, making orchestration platforms an increasingly important category," said Katie Norton, Research Manager, IDC. "In more autonomous delivery models, the ability to coordinate agents across the software lifecycle while maintaining clear boundaries and visibility becomes a key requirement. GitLab's approach aligns with this shift by focusing on policy enforcement and governance alongside efforts to reduce operational friction," said Norton.

GitLab said GitLab Dedicated availability will follow during the GitLab 18.8 release cycle.