GM redesigned its engineering workflows around AI agents — and tripled its merged pull requests

GM’s autonomous-vehicle unit says engineers spend only about 15% of their time coding, so it shifted more workflow work to AI agents.

The agents handle data analysis, triage, testing, and experimentation by connecting to internal tools, vehicle data, and MCP servers.

GM says the approach has tripled merged pull requests, sped up releases, and reduced defects, pointing to gains beyond coding copilots.

The company argues that redesigning engineering workflows around AI agents could become a broader productivity model for software teams.