GitLab is evolving into an AI-native platform for software engineering, aiming to foster human-AI collaboration. This transformation occurs across three key layers: its unified data platform, its software control plane, and its user experience. The platform acts as a system of record, securely holding valuable digital assets and unstructured data within the GitLab environment. It also serves as a software control plane, orchestrating critical business processes through Git repositories and APIs. GitLab provides an integrated interface to reduce context-switching for its millions of users. The new GitLab Duo Agent Platform integrates and extends these three layers, offering extensibility and interoperability. A Knowledge Graph indexes code and unstructured data for optimized agent access, accelerating reasoning and delivering better outcomes. An Orchestration Layer is being added to the control plane, enabling agent registration and multi-agent flows. GitLab is also introducing native agents for every development stage, facilitating seamless human-AI collaboration. Release 18.3 expands integrations, interoperability, and context awareness across the entire software development lifecycle. This includes a Model Context Protocol server for universal AI integration and CLI agent support for various AI providers. Agentic Chat support is now available in Visual Studio IDE and GitLab UI for all premium customers. New automated development flows, such as Issue to MR Flow and Convert CI File Flow, are introduced to handle mundane tasks. The Knowledge Graph provides real-time code intelligence for faster, more accurate search results. Enterprise governance features like Agent Insights offer transparency into agent decision-making, addressing security and compliance concerns.