General Motors has announced plans to launch its own proprietary in-vehicle AI assistant later this year, a system designed to go far beyond what Google's Gemini can accomplish inside a car. The new assistant will be deeply integrated with OnStar, GM's telemetry systems, and the company's proprietary vehicle knowledge, enabling predictive maintenance, real-time diagnostics, and auto-specific commands that a general-purpose assistant simply cannot handle. GM is working with an unnamed large language model provider on this technology, signaling a deliberate effort to diversify its AI partnerships while retaining control over the most valuable data generated by its vehicles.
Key facts at a glance
- GM will launch its own in-vehicle AI assistant later this year.
- The system integrates with OnStar, telemetry, and predictive maintenance.
- It complements, rather than replaces, Google's Gemini assistant.
- GM's assistant will access proprietary vehicle data that Gemini cannot reach.
- The technology is being built with an unnamed large language model provider.
- GM's move reflects concerns about single-vendor AI dependency and data ownership.
The limits of a dashboard AI
Google's Gemini launched in eligible 2022 and newer GM vehicles earlier this year, handling conversational queries, temperature adjustments, and radio controls. But according to Anna Santos, GM's director of product management for voice and AI, there is a hard ceiling on what a top-level AI can achieve when it lacks access to the vehicle's inner workings. “There's a limit to what an AI that's just sort of sitting at the top level of the vehicle can do,” Santos said in a recent interview. That limitation has driven GM to develop a parallel system that dives deep into the car's mechanical state, driving patterns, and owner preferences.
The new assistant will understand not just what the driver asks, but what the vehicle itself is experiencing. It will be able to anticipate problems before they occur, provide real-time diagnostics, and execute complex multi-step commands that require awareness of the car's overall status. Santos offered a concrete example: a “kids setting” command that would simultaneously adjust music, seats, climate, and door locks. A general-purpose assistant could potentially handle some of these tasks individually, but it would not understand the context of a family trip or the vehicle's safety systems the way a purpose-built AI could.
Data ownership as a strategic imperative
The distinction between GM's assistant and Gemini is not just technical; it is commercial. Gemini operates like a phone assistant projected onto the dashboard, relying on cloud-based knowledge and generic user inputs. GM's proprietary system, by contrast, draws on data that belongs to GM, not Google. “It's data that's going to be proprietary to GM,” Santos emphasized. This is a critical point in an era where data is often more valuable than hardware. Every mile driven, every diagnostic code, every braking pattern is a piece of information that can improve the driving experience, enable predictive maintenance, and ultimately feed into the development of autonomous vehicles.
Google has been embedding AI across every product it controls, from search to cloud to automotive. By building a parallel AI system, GM is effectively drawing a line: the dashboard may be shared, but the vehicle's intelligence—and the data that powers it—remains in GM's hands. This echoes concerns raised by Microsoft CEO Satya Nadella, who has warned companies about outsourcing their thinking to a single AI provider. In the automotive industry, where safety, reliability, and customer trust are paramount, such dependency is risky. If a single AI provider controls both the general assistant and the vehicle-specific intelligence, the automaker loses leverage and the ability to differentiate.
Two assistants, two roles
GM has made clear that it is not replacing Gemini. Both assistants will coexist, each with a defined role. The Gemini layer will continue to handle general queries, entertainment, navigation, and other tasks that rely on broad world knowledge. The GM layer will handle everything the car knows about itself—mechanical condition, battery state, tire pressure, recommended service intervals, and even the driver's habits and preferences. This division of labor allows GM to offer the best of both worlds: the conversational fluency of a major AI model and the deep, vehicle-specific intelligence that only an automaker can provide.
The coexistence also has a practical benefit. Drivers who are already accustomed to Gemini's interface will not be forced to adapt to an entirely new system. Instead, the GM assistant will quietly operate in the background, ready to step in when the conversation turns to the vehicle itself. For example, a driver might ask Gemini about the weather forecast, then ask the GM assistant about the check-engine light that just appeared. The transition could be seamless, with both systems working together to provide a unified experience.
Feeding the autonomous vehicle pipeline
Beyond the immediate benefits to drivers, GM's new AI assistant is a strategic component of the company's broader push toward autonomous driving. GM has been aggressively advancing its self-driving capabilities, with a focus on robotaxi-readiness. An in-vehicle AI that understands predictive maintenance and telemetry feeds directly into fleet management for autonomous operations. When a fleet of driverless vehicles is on the road, operators need to know not just where each car is, but how it is performing mechanically. The AI assistant can monitor the health of every vehicle in real time, scheduling maintenance proactively and minimizing downtime.
In this sense, the assistant is not just a consumer feature; it is a critical tool for the future of mobility. Santos called it “the beginning of a broader AI journey,” suggesting that GM's ambitions extend far beyond voice commands and climate control. The company is building an intel infrastructure that will underpin everything from predictive maintenance to fleet optimization to personalized in-car experiences. As autonomous vehicles become more common, the ability to process vehicle data in real time will be a significant competitive advantage.
The unnamed LLM provider and the vendor diversification strategy
GM has been tight-lipped about the large language model provider powering its new assistant. The fact that the company is keeping the name under wraps is telling. It suggests that GM is keeping its options open, evaluating multiple AI models and avoiding the single-vendor dependency that its Gemini partnership already created. By working with an unnamed provider, GM can negotiate from a position of strength, potentially shifting between providers as the technology evolves. This is a smart move in a market where AI models are rapidly improving and nobody wants to be locked into a system that becomes outdated.
The decision to build a proprietary assistant also positions GM as a player in the larger AI ecosystem, rather than a mere customer of Big Tech. As cars become increasingly software-defined, the automaker's relationship with AI providers will determine not only the user experience but also the revenue streams. Data from GM's vehicles could be valuable for training models, improving autonomous systems, or even offering new services to drivers. By keeping the data in-house, GM retains the ability to monetize it directly, rather than handing it over to a third-party AI company.
The evolution of OnStar and vehicle intelligence
GM's new assistant builds on a long history of in-vehicle connectivity. OnStar, launched in 1996, was one of the first embedded telematics systems in the industry, offering emergency services, navigation, and remote diagnostics. Over the years, OnStar has evolved to include mobile apps, Wi-Fi hotspots, and advanced safety features. The new AI assistant takes that evolution to the next level, transforming OnStar from a reactive service into a proactive, intelligent companion that understands the vehicle and its driver on a deeper level.
This is part of a broader trend in the automotive industry, where automakers are increasingly developing their own software stacks to compete with tech giants. Ford, Mercedes-Benz, and BMW have all made significant investments in proprietary AI systems, recognizing that the driver experience is no longer defined solely by horsepower and interior materials, but by the quality of the digital interface. GM's approach stands out because of its emphasis on data ownership and its willingness to partner with multiple AI providers rather than going all-in with one company.
The automotive AI landscape is still in its early days, and there is no clear winner yet. Some automakers have chosen to embed Amazon's Alexa deeply into their vehicles, while others, like Tesla, have built entirely vertical AI systems from scratch. GM's strategy of using a general-purpose AI as a complement to a proprietary system is a hybrid approach that may prove to be the most practical. It allows the company to offer leading-edge AI capabilities without sacrificing control over the data and intelligence that will define the next generation of vehicles.
As GM prepares to launch its assistant later this year, the industry will be watching closely. The success of the system will depend not only on the quality of the AI model but also on how seamlessly it integrates with the vehicle's hardware, software, and cloud services. GM's years of experience with OnStar and its investment in telematics give it a solid foundation, but the challenge is significant. The company must ensure that the assistant is reliable, secure, and genuinely useful to a wide range of drivers. If it succeeds, GM could set a new standard for what an in-vehicle AI should be.