GitLab + AWS Bedrock
- GitLab announced integration with AWS Bedrock to run model inference within GitLab CI using customer AWS accounts. - The integration supports bring‑your‑own‑model, per‑request billing, and IAM‑based governance to keep inference in customer control. - The hookup lets teams embed model calls into CI/CD while retaining AWS security and billing boundaries (x.com).
GitLab said on April 21 that customers can now run GitLab Duo Agent Platform inference through Amazon Bedrock inside their own Amazon Web Services accounts. (about.gitlab.com) The setup routes model calls from GitLab into Bedrock models the customer already has access to, using existing Identity and Access Management policies and existing AWS spending commitments. GitLab said credits bought through AWS Marketplace also count toward those commitments. (about.gitlab.com) Amazon Bedrock is Amazon’s managed service for calling large language models without hosting the models yourself. GitLab’s role here is the orchestration layer: developers trigger AI features in GitLab, and the inference runs against models and billing boundaries controlled in AWS. (about.gitlab.com) GitLab has been building that self-hosted path for more than a year. Its GitLab Duo Self-Hosted offering reached general availability in 2025 for code suggestions and chat, with support for private-cloud and secure-cloud deployments including AWS Bedrock. (about.gitlab.com) The new link pushes that model deeper into software delivery work, where teams want AI inside the same pipelines that build, test, and ship code. GitLab’s documentation says self-managed customers can use a self-hosted AI Gateway so request and response data stay in their own environment while connecting to Bedrock as the model backend. (docs.gitlab.com) GitLab’s Bedrock deployment guide shows the plumbing in practical terms: a self-managed GitLab instance, an AI Gateway, and Bedrock as the large language model provider. The company recommends Bedrock access in a target AWS region and documents model assignment, transport security, and verification steps. (docs.gitlab.com) GitLab also says customers can mix model providers inside one instance. Its supported-platform documentation gives Bedrock and Azure OpenAI as examples, with different models assigned to different GitLab features. (docs.gitlab.com) This AWS tie-up extends a broader GitLab push to sell AI as part of its DevSecOps platform rather than as a separate chatbot. In April 2025, GitLab launched general availability of GitLab Duo with Amazon Q for Ultimate self-managed customers on AWS. (about.gitlab.com) Investors treated the April 21 announcement as material news. Market coverage from Tuesday said GitLab shares rose more than 6% in extended trading after the company announced the deeper AWS collaboration. (msn.com) The pitch to customers is straightforward: keep GitLab as the place where developers work, keep Bedrock as the place where models run, and keep security, governance, and cloud spend inside the customer’s own AWS account. (finance.yahoo.com)
