GitLab brings agentic AI to single-tenant SaaS for high-security enterprises
New updates to GitLab Duo and version 19.3 integrate autonomous AI agents into isolated infrastructure to meet strict data residency requirements.
GitLab has integrated its Duo Agent Platform into its Dedicated single-tenant SaaS infrastructure, allowing enterprises to run agentic AI workloads within their own security boundaries. The move targets regulated industries where data residency and security isolation are primary barriers to AI adoption.
As part of this rollout, the AI Gateway for the GitLab Duo Agent Platform now operates inside GitLab Dedicated. This ensures that any data processed by AI remains within the customer's existing security perimeter. Alongside this infrastructure shift, GitLab 19.3 introduces several autonomous capabilities, including "Agentic SAST Vulnerability Resolution," which enables development teams to generate ready-to-merge fixes for multiple vulnerabilities in a backlog simultaneously.
To further streamline automation, the company launched the Flow Creator Agent. This tool allows users to describe desired automations in plain language to generate runnable flows, effectively removing the requirement for manual schema mapping. Additionally, GitLab has introduced a Secrets Manager—currently in limited availability as a paid add-on—that supports custom tools as well as Terraform, OpenTofu, and Kubernetes. To help organizations control costs associated with these autonomous workloads, GitLab Credits usage caps are now generally available for managing monthly spending.
The shift to autonomous orchestration
These updates signal a transition from generative AI, which primarily suggests code, to agentic AI, which can execute complex workflows and remediate security flaws autonomously. By combining this autonomy with a Secrets Manager and isolated infrastructure, GitLab is attempting to make autonomous agents safe for Fortune 100 companies. The goal is to ensure that AI agents operate under the same strict permission and residency models as human developers, reducing the risk of data leakage or unauthorized access during automated remediation.
Manav Khurana, chief product and marketing officer at GitLab, stated that these updates extend the speed and control required by the regulated and data-sensitive segments of the enterprise market. According to Khurana, every new capability—from the location where an agent runs to the secrets it can access—is designed to extend control into the trusted software delivery workflows that enterprises already rely upon.
Future of DevSecOps
GitLab is positioning itself as an intelligent orchestration platform for DevSecOps, moving beyond simple tool integration toward a model where AI manages the lifecycle of software delivery. The industry will now be watching to see if the removal of residency barriers leads to a measurable increase in the adoption of autonomous vulnerability patching among highly regulated firms. While the Secrets Manager is currently in limited availability, its broader rollout will be a key indicator of how GitLab intends to scale the trust model for agentic AI.