Rivian recently announced an AI assistant that allows drivers and passengers to interact with its electric vehicles simply by speaking. But RJ Scaringe, the automaker’s founder and CEO, told TechCrunch Disrupt this week that in the future, Rivian cars will be able to run AI agents made by other companies like Anthropic.
“That’s where it becomes magic,” Scaringe said on stage.
Scaringe did not say when Rivian would implement this, and the company declined to provide details, instead pointing to an event it will host in December called AI & Autonomy Day.
But it’s not hard to imagine how Rivian might integrate Claude, OpenAI’s Dots, Meta’s Muse assistant, etc. into its software. Rather than relying on Apple CarPlay or Android Auto, the company developed its in-car software from the ground up, building many of its own apps like navigation (though it has also worked on integrating popular in-car apps like Apple Music).
Rivian says it has created its own in-house agent framework that understands what action a person wants to take when they talk to a vehicle and routes it to the appropriate part of the vehicle software. In most cases, this includes asking the car to lower the temperature, turn on the seat warmers, or summarize the news.
“Think of it like a traffic light: You ask the vehicle something, and the vehicle decides which model to use. It can use local models running at the edge and on the vehicle. For more complex questions, you can send it to the cloud. And even then, it uses a different model,” Scaringe said.
As of May of this year, agents had access to external apps like Google Calendar. And Scaringe told Disrupt that the next step will be adding products from cutting-edge AI companies, but getting them to work inside the car won’t be easy.
“It takes a lot of plumbing behind the scenes to make sure it works, and it’s not just within Rivian. It also takes a lot of coordination with the lab to make sure that we can do that plumbing,” he said, adding that having this kind of software near safety-critical systems means Rivian has to be very careful.
“You don’t want the agent to decide, ‘It’s time to put on the brakes, or it’s time to take control of the car,’ so everything about vehicle-level control is not being handed to the agent,” he said.
Rather, he believes this implementation will be an extension of how people currently use these agent apps. That means Rivian will have to find a way to connect its vehicles to all the context that apps like Muse collect from users.
For example, Scaringe said drivers can ask Rivian’s assistant to turn on the ventilated seats before going inside. “Then at the same time you say, ‘We talked about this meal yesterday. We’re having guests over on Friday. Can you help us figure out what to make?'” he said. The car can then derive the driver’s context from conversations the driver has had with the assistant on another device.
Of course, Rivian isn’t the first American car company to add AI agents to its cars’ software. Earlier this year, Tesla added the ability for users to talk to both SpaceX’s Grok chatbot and the Grok Bot assistant.
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
