AI agents are coming to Rivian’s cars
TechCrunch Sean O'Kane
Rivian says its cars will eventually run AI agents from companies like Anthropic. The risky part: it wants the smarts without letting software touch the brakes or steering.
Based on reporting by TechCrunch, Sean O'Kane — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Rivian is already letting people talk to its electric vehicles, and now the company says that’s only the first step. At TechCrunch Disrupt, CEO RJ Scaringe said Rivian wants future cars to run AI agents made by outside companies such as Anthropic. He called that the point where things get “magical.”
The company isn’t saying when that happens. Rivian pointed to its AI & Autonomy Day in December and stopped there. But the outline is pretty clear: Rivian wants its cars to behave less like locked-down gadgets and more like software platforms that can tap different models depending on the task.
That matters because Rivian built its in-car software itself instead of leaning on Apple CarPlay or Android Auto. It also made a lot of its own apps, including navigation, while still integrating services like Apple Music. Rivian says its own agentic framework already figures out what a driver wants and sends the request to the right part of the vehicle software. That can be simple stuff, like lowering the temperature, turning on seat warmers, or summarizing the news.
Scaringe said the system works a bit like a traffic light. Simple requests can stay on a local model running inside the vehicle; more complex ones can go to the cloud, with Rivian choosing among different models. By May, the agent could also reach external apps such as Google Calendar. The harder next step is making frontier AI agents work in the car at all, which means a lot of plumbing and coordination with the labs behind them.
Rivian is drawing a hard line around safety-critical functions. No agent gets to decide when to brake or steer. Instead, Scaringe imagines something closer to a cross-device assistant that can use context from elsewhere, like a conversation a driver had on another device. He gave the example of someone asking the car to turn on the ventilated seats, then continuing a dinner-planning conversation about guests coming on Friday.
Rivian wouldn’t be the first American automaker to move here. Tesla already added support for talking to SpaceX’s Grok chatbot and Grok Bot assistant earlier this year. But Rivian’s version sounds more ambitious, and more careful.
My take — AI-written commentary, not fact-checked reporting
This is the right kind of AI in a car: useful until it gets anywhere near the controls. The industry keeps trying to turn every dashboard into a demo stage, when most drivers would probably settle for software that answers the question, then stays out of the way. That’s not less ambitious. It’s just less ridiculous.
Read more about this at: TechCrunch