GitLab wraps coding agents in compliance

GitLab said Tuesday it is wiring its coding agents into Google Cloud’s Vertex AI while keeping those agents inside GitLab’s own compliance controls. (about.gitlab.com) The company said GitLab Duo Agent Platform can now call foundation models through Vertex AI, including Google’s Gemini models, and enterprises can count that usage against existing Google Cloud spending commitments. (about.gitlab.com) GitLab also said organizations can run its artificial intelligence gateway on Google Cloud runtimes such as Google Kubernetes Engine and Cloud Run, instead of standing up a separate stack just for agent workloads. (markets.financialcontent.com) Coding agents are software tools that do more than suggest code: GitLab’s platform is designed to let specialized agents plan work, write code, review changes, and help remediate security issues across the software delivery process. (about.gitlab.com) GitLab made Duo Agent Platform generally available on January 15, 2026, for Premium and Ultimate customers on GitLab.com and self-managed deployments, after pitching the product as a way to extend artificial intelligence beyond code generation into the rest of DevSecOps. (about.gitlab.com) Vertex AI is Google Cloud’s managed platform for building and using generative artificial intelligence, and Google says it gives customers access to more than 200 foundation models alongside its own Gemini family. (cloud.google.com) That pairing targets a problem large companies keep raising with coding agents: they want access to strong models, but they also want logs, policy controls, and audit trails around what autonomous tools do inside production software pipelines. GitLab said agent actions in this setup remain governed by its built-in compliance and audit features. (about.gitlab.com) Google has been pushing the same enterprise pitch around agents. Its Vertex AI Agent Builder says companies need ways to build, scale, and govern agents that are tied to enterprise data and workflows, not isolated chat tools. (cloud.google.com) GitLab’s announcement does not say customers must use only Google models. For self-hosted deployments, the company said its “Bring Your Own Model” option still lets teams connect approved models and gateways of their choice. (markets.financialcontent.com) The immediate change is not that coding agents suddenly exist. It is that GitLab is trying to make them fit the procurement, infrastructure, and audit rules that already govern enterprise software delivery. (about.gitlab.com)