GM builds its own in-vehicle AI assistant to access vehicle data beyond Google Gemini

General Motors will launch its own in-vehicle AI assistant later this year, built to access vehicle data that Google's Gemini cannot reach. The system integrates with OnStar, telemetry, and proprietary vehicle knowledge to offer predictive maintenance, real-time diagnostics, and auto-specific commands. For AI builders, this signals a shift toward multi-model, data-owned strategies in automotive AI.

What happened

GM is developing a native AI assistant that operates alongside Google Gemini. The GM layer targets the car's mechanical state, driving patterns, and owner preferences, while Gemini handles general queries and entertainment. GM is working with an unnamed large language model provider, indicating a multi-model approach that avoids single-vendor dependency. The company plans to launch the proprietary system later this year.

Why AI builders should care

The GM strategy demonstrates how automakers can decouple vehicle intelligence from consumer-facing assistants to preserve data sovereignty. A multi-model approach with vehicle-specific AI enables predictive maintenance and fleet telemetry integration, expanding AI use cases beyond general consumer tasks. GM's decision to build a parallel system rather than hand all vehicle intelligence to Gemini reflects a broader concern about outsourcing critical system data to a single AI provider.

Practical implications

A native GM AI layer will coexist with Gemini, offering vehicle-centric capabilities. The architecture implies data streams from OnStar, telemetry, and vehicle diagnostics feeding into an AI assistant tailored to individual car knowledge and real-time state. For example, a 'kids setting' command adjusts music, seats, climate, and door locks simultaneously. This vehicle-specific knowledge enables predictive maintenance and real-time diagnostics that a general-purpose assistant cannot provide.

Caveats

Details about rollout timing, the exact model powering the GM AI, and the set of vehicles supported are not finalized. The core evidence comes from a single article and secondary sources; specifics may evolve as GM formalizes plans. The unnamed LLM provider suggests GM is keeping options open, but the actual model and integration details remain undisclosed.