Rivian’s chief executive RJ Scaringe used his appearance at TechCrunch Disrupt to outline a future in which its electric vehicles run AI agents built by outside companies, including Anthropic, though he set no timetable and the carmaker declined to offer further detail beyond pointing to an upcoming product event. ‘That’s where it becomes magical,’ Scaringe said onstage.
The announcement extends an AI push that Rivian has been building in-house. The company already operates what it describes as a proprietary agentic framework: a routing system that takes a driver’s spoken request and directs it to the appropriate part of the car’s software. As Scaringe explained it: ‘Think of it as like a traffic light. So if you ask the vehicle something, it decides what model it’s going to use. It can use a local model that’s running on the edge and the vehicle. If it’s a more complex question, it can go up to the cloud. And then even for that, we have different models that we use.’
Built from the ground up
Rivian has developed its in-car software independently rather than relying on Apple CarPlay or Android Auto, constructing its own applications (including navigation) while integrating selected third-party services such as Apple Music. That vertical approach is what makes the idea of plugging in frontier AI models plausible, though not straightforward.
In May 2026, Rivian’s assistant gained access to Google Calendar as its first external agentic integration, according to the company’s own announcement. Scaringe described the addition of models from companies such as Anthropic, or products like OpenAI’s Dots or Meta’s Muse, as the logical next step, but stressed the work involved. ‘There’s a lot of plumbing behind the scenes we have to do to make sure that that works, and it’s not just within Rivian. That’s also a lot of coordination with the labs to make sure they enable that plumbing,’ he said.
Rivian’s 10-Q filed with the SEC for the period ended 30 June 2026 describes the Rivian Assistant as powered by a proprietary, multi-modal AI foundation and an in-house agentic framework, and states that the feature launched with Connect+ on all R1 vehicles, with a plan to bring it to R2 vehicles via an over-the-air update later in 2026.
Safety limits and the road ahead
Scaringe was direct about where third-party agents will not be permitted to reach. ‘You wouldn’t want an agent to determine, “hey, it’s time to apply the brakes, or time to steer the vehicle.” So anything that’s vehicle-level controls are not being handed to an agent,’ he said, drawing a firm line between in-cabin convenience features and safety-critical systems.
The intended use cases are more in line with how consumers already use AI assistants on their phones. Scaringe sketched out a scenario in which a driver asks the car to activate ventilated seats and, in the same breath, picks up a conversation from another device about planning a meal for upcoming guests, drawing on context from an earlier discussion held on a separate device. Making that work would require Rivian to connect its vehicles to the personal context that apps like Muse collect on their users, itself a substantial integration challenge, by Scaringe’s own account.
Rivian would not be the first American car company to add outside AI agents to its software. Earlier in 2026, Tesla added a feature allowing users to talk to SpaceX’s Grok chatbot and its Grok Bot assistant.
The company pointed journalists to its ‘AI & Autonomy Day’ event, scheduled for December, for further detail. In December 2025, CNBC reported that Rivian held its first such event at its Palo Alto offices, where it announced a custom chip, new car computer and AI models intended to underpin autonomous features on future vehicles. Scaringe gave no equivalent level of detail at Disrupt, and the company declined to elaborate ahead of the December event.
