Rivian has introduced an AI assistant, enabling vocal interaction with its electric vehicles. At TechCrunch Disrupt, CEO RJ Scaringe revealed plans for future Rivian vehicles to support third-party AI agents, such as those from Anthropic.
Scaringe enthusiastically described this potential integration as “magical”.
The timeline for this implementation was not disclosed by Scaringe, and Rivian chose not to elaborate, instead referring to its upcoming “AI & Autonomy Day” in December for further announcements.
Visualizing the integration of external AI assistants like Anthropic’s Claude, OpenAI’s Dots, or Meta’s Muse into Rivian’s platform is straightforward, particularly since Rivian developed its in-car software independently, bypassing Apple CarPlay or Android Auto, and crafted many core applications, including navigation (while also incorporating services like Apple Music).
Rivian asserts it has engineered an internal agentic framework designed to interpret driver commands and direct them to the appropriate vehicle software components. Typically, these interactions involve standard requests such as adjusting temperature, engaging seat heaters, or providing news summaries.
Scaringe likened this system to a “traffic light,” explaining that upon receiving a query, the vehicle determines the optimal AI model to employ. This could be a local, edge-based model for simpler tasks, or a cloud-based model for more intricate questions, with a variety of models available for the latter.
By May, the AI agent was already capable of integrating with external applications like Google Calendar. Scaringe indicated at Disrupt that the next evolutionary step involves incorporating offerings from leading AI firms, acknowledging the technical challenges this presents for in-car functionality.
He emphasized the extensive backend development and inter-company coordination required to facilitate these integrations. Scaringe also highlighted the imperative for extreme caution when embedding such software close to safety-critical vehicle systems.
Scaringe clarified that agents would not control critical vehicle functions like braking or steering, ensuring that “vehicle-level controls are not being handed to an agent”.
He foresees this implementation as an extension of existing agentic app usage, requiring Rivian to bridge the vehicle with the contextual data gathered by applications like Muse from their users.
Scaringe provided an example: a driver might first command the Rivian assistant to activate ventilated seats, then immediately inquire, “We discussed this meal yesterday; I’m hosting guests Friday. Could you help me plan the menu?” The vehicle could then access the driver’s contextual conversation history with the assistant from a different device.
Rivian is not pioneering the integration of AI agents into automotive software among American car manufacturers; Tesla previously introduced a feature enabling user interaction with both SpaceX’s Grok chatbot and its Grok Bot assistant.