gitlab duo agent platform

Overview

Duo Chat is a conversational AI coding assistant developed by GitLab, the leading source code management platform. Designed to interact through natural language, it offers real-time guidance to both engineering and non-engineering users throughout the entire software development lifecycle. By supporting every stage of development, it helps teams accelerate code comprehension, enhance collaboration, and become proficient with GitLab in record time.

With Duo Chat, developers can explain unfamiliar code, refactor existing code, generate test suites, and receive intelligent code suggestions that boost both productivity and code quality. Privacy is a core priority: GitLab does not use customer data to train AI models, and organizations can control AI access to sensitive information. Duo Chat is embedded within the GitLab interface and also integrates seamlessly with popular IDEs such as VS Code and the JetBrains suite, making it a versatile companion for modern development teams.

Core capabilities

AI Orchestration Across the SDLC

Purpose-built agents for planning, coding, security analysis, analytics, and more, each using GitLab context to deliver precise, relevant results that support real work.

AI Catalog

A centralized catalog that gives teams a single place to explore, activate, and manage agents and flows, making automation simple to adopt and consistent to operationalize across projects and groups.

Customizable Agents and External Integration

Create or refine agents to follow your conventions and quality expectations, and integrate external agents such as Claude Code and Codex into the GitLab workflow.

Agentic Workflows (Flows)

Combine one or more agents into guided, multi-step sequences that automate manual steps end to end, with event-driven triggers from GitLab events and reliable, repeatable execution under your rules and identity.

Enterprise-Grade Governance and Control

Policy-driven AI control determines which agents and flows are allowed, where they operate, and which models they use, aligned with existing roles and group structure; full traceability includes trigger details and CI/CD job logs; self-hosted LLMs are supported for self-managed deployments.

Extensibility and Model Flexibility

Connect internal systems, third-party tools, and external AI services; agents and flows can be updated, extended, or replaced over time; select the right LLM for agents, flows, and agentic chat.

GitLab Orbit Context Graph

Powered by GitLab Orbit, a lifecycle context graph that connects code, pipelines, and deployments as grounded context for every agent, improving agent speed, accuracy, and cost-effectiveness.

Usage-Based Pricing with GitLab Credits

GitLab Credits are pooled across the organization and cover both synchronous agent interactions and asynchronous agentic flows, enabling AI usage that scales with real usage instead of seat count.

Data Privacy and Transparency

GitLab does not train generative AI models on private data, and outputs generated by GitLab Duo can be used at your discretion, with GitLab providing defense against third-party claims arising from output use; governance and transparency details are available via the AI Transparency Center.

Broad Use-Case Coverage

Pre-built agents and flows address planning, security analysis, automated code reviews, CI/CD pipeline diagnosis and fix, pipeline migration to GitLab CI/CD, issue-to-merge-request transformation, code generation, refactoring, and custom agent/flow creation.

Use cases

Automatically review newly submitted merge requests and flag potential issues

Automatically recommend fix code snippets based on test failure logs

Monitor production environment logs and trigger automatic rollback operations

Generate standardized CI/CD configuration templates for new projects

Automatically add necessary comments and documentation to code

Periodically scan dependency libraries and create upgrade issues

Help new members quickly understand project architecture and coding standards