Rivian plans to integrate third-party AI agents into its vehicle software. CEO RJ Scaringe announced that the company aims to support models from developers like Anthropic in the future.
The company has developed an internal agentic framework that processes voice commands and routes them to specific vehicle functions. This system currently handles tasks like climate control, seat heating, and news summaries.
The software architecture uses a hybrid approach, utilizing local edge models for simple requests and cloud-based models for more complex queries. As of May, the system gained access to external services such as Google Calendar.
Rivian intends to maintain strict separation between AI agents and safety-critical systems. Scaringe emphasized that vehicle-level controls, such as steering or braking, will not be accessible to these agents.
The company plans to share further details during its AI and Autonomy Day event scheduled for December. Rivian is currently working on the technical infrastructure required to integrate external AI products into its ecosystem.
