Home / Technology / GM Reboots Engineering with AI Agents
GM Reboots Engineering with AI Agents
10 Jul
Summary
- GM's AI initiative increased engineer pull requests by 300%.
- GM is re-architecting its systems for AI agent interaction.
- A three-layer control plane manages AI task execution.
General Motors has embarked on a significant re-architecture of its technology infrastructure to better support AI agents in engineering tasks. Recognizing that traditional software ecosystems optimized for human operators create friction for AI, GM's tech team is enabling direct agent access through open APIs and developing custom Model Context Protocols (MCPs).
This strategic shift aims to reduce barriers for agentic interaction with industry tools and GM's proprietary applications. A key element of this new operational backbone is a sophisticated three-layer control plane. This system is orchestrated by a business-aware LLM, supported by task-specific LLMs, and utilizes MCPs to interface with tools and data.
Approximately eighteen months into this ambitious program, GM is observing remarkable efficiency gains. The re-architected system has successfully driven a 300% increase in merged pull requests. This indicates that AI agents are now acting as significant force-multipliers for GM's engineering teams, accelerating development and innovation.
Rashed Haq, VP of autonomous vehicles at GM, is scheduled to present further technical details and insights on this transformation at the VB Transform 2026 conference on July 14. His session will explore how GM is redesigning its agentic systems to achieve breakthrough engineering velocity.
