Case Study

GitLab powers stable, secure software development with Claude

Gitlab + Anthropic logo lockups

GitLab, the most comprehensive DevSecOps platform, uses Claude to drive AI-powered features across their software development, deployment and security platform. By leveraging Claude's capabilities, GitLab delivers innovative AI solutions while maintaining the stability and security its customers demand.

  • Several Claude-powered features are implemented across the platform
  • Claude delivers high-context and accurate code generation for users
  • Accelerates development and deployment of AI features for Gitlab engineering team

Balancing innovation with stability in DevSecOps

GitLab’s AI-powered platform helps companies manage the entire software development lifecycle. With security and speed at the core of its value proposition, GitLab needed to balance AI’s potential with taking a privacy-first approach to embedding AI capabilities across the platform.

"Our mission is to enable everyone to contribute to and co-create the software that powers our world," says Taylor McCaslin, Group Manager of Product, Data Science AI/ML at GitLab. "That's a major undertaking we do not take lightly."

As GitLab embarked on its journey to integrate generative AI, they recognized that collaborating with an AI company like Anthropic would be the most effective way to integrate the technology into their platform. GitLab Principal Engineer Jessie Young explains, "As a company that specializes in APIs, we wanted to weave AI into various features without reinventing the wheel. Partnering with Anthropic allowed us to access powerful and easy-to-use AI models that could seamlessly integrate with our platform."

Implementing a multi-model approach with Claude

GitLab hit its stride when more advanced AI models became available. The company has now integrated AI throughout its platform, with features for generative coding, interactive chat, planning summarization, and explaining and remediating software vulnerabilities. To ensure they're using the best tool for each task, GitLab created a model evaluation team that assesses new vendors and new versions of models.

"Since we have AI-powered use cases across the entire software development lifecycle, we need an approach that enables us to choose the right model for the right use case. This makes the Claude model family approach a huge advantage for our team,” says McCaslin.

GitLab chose Claude 3 models to power their AI-powered Duo features due to their performance and the fully featured developer experience. "We see consistently strong performance from Claude 3 models for thoughtful, holistic, and contextualized code generation and other software development-related tasks," McCaslin notes, attributing this partly to Claude's long context window.

Gitlab product mockup

Empowering engineers and ensuring reliability

For GitLab, Anthropic's intuitive developer tools have enabled their team to build AI-powered features efficiently.

"The tooling Anthropic provides us is approachable for somebody who doesn't have a machine learning background," says Young. "Anthropic's developer tools and docs enable us to have dozens of people with more traditional software engineering backgrounds work on AI features."

GitLab values Anthropic's commitment to transparency and privacy in this rapidly evolving AI landscape. This reliability is essential for GitLab as they work to deliver stable, secure software development tools to their customers.

Anthropic shares our values of privacy and transparency and are straightforward to work with. That's especially important for us as an enterprise organization that builds reliable software.

— Taylor McCaslin, Group Manager of Product, Data Science AI/ML

Looking to an AI-powered future

As GitLab continues to evolve their platform, the company is well-positioned to lead in AI-driven DevSecOps. With Claude powering key Duo features and providing a foundation for exploring future use cases, GitLab can focus on delivering cutting-edge capabilities that their customers can rely on.

By choosing Claude, GitLab has found foundation models that provide powerful AI capabilities and align with their commitment to stability and security in software development.