Inside Rivian's 'all-out' push towards AI-defined vehicles

Everything from the software to the computers or brains in Rivian vehicles is developed and tested within the premises of 607 Hansen Way in Palo Alto.

Built in the 1950s, the building's first tenant was Varian Associates, widely known as one of Palo Alto's first technology companies.

The common lore behind 607 Hansen Way is that it was built with potential alternative uses in mind, such as an educational facility, which may explain the building's resemblance to an elementary school.

Philbin is a veteran of the robotaxis industry, previously serving as senior director at Amazon-backed company Zoox and director of engineering at Waymo.

He was hired by Rivian in 2022 to lead as Vice President of Autonomous Driving and AI. This is a key indicator that the company's shift to an AI-centric approach to self-driving didn't happen overnight.

Philbin told Business Insider that since joining the company, the team has grown to include employees who previously worked on fully autonomous driving systems, advanced driver assistance systems, and machine learning projects.

Philbin said he manages a team of about 300 people.

“We all share a passion for the product and actually delivering the functionality directly to the customer,” he said.

Rivian's philosophy of owning every part of the software stack is led by Chief Software Officer Bensaid. He joined the company in 2019.

In addition to co-leading the joint venture between Rivian and Volkswagen, Benside also serves as spokesperson for Rivian's in-vehicle AI assistant.

Benside told Business Insider he believes there are major flaws in the interface between drivers and vehicles.

“I think the fact that we're touching screens, or even using buttons in some cars today, is an anomaly. It's a bug, not a feature,” he said.

His vision is a hands-free, fully voice-controlled interface.

Bensaid demonstrated Rivian Assistant at Autonomy & AI Day, showing how drivers can use the AI ​​assistant to control their Google Calendar and send messages to their contacts.

“This assistant is more than just a chatbot integrated on top of the vehicle UI,” Bensaid said during his presentation.

Rajagopalan leads the team responsible for the hardware side of Rivian's onboard computing and electrical architecture. Previously, he was a senior director at Tesla, where he was involved in the development of the Model 3.

Mr. Rajagopalan led Rivian's transition to a “zone architecture” design. This essentially means more vehicle functions are controlled by fewer computers. One benefit of this approach is that Rivian's second-generation vehicles were built with less wiring, thereby reducing costs.

Central to Rivian's self-driving roadmap is the company's decision to design its own chips, similar to what Tesla did in 2019.

The first-generation Rivian Autonomy Processor (RAP1) chip provides the computing power to process all the data collected by Rivian's sensors for autonomous driving.

Rajagopalan said Rivian began considering the idea for its chip about three and a half years ago, starting with a feasibility and business case analysis before the design process began.

“The lead time from building silicon to actual deployment is very long,” she says. “So, in a way, it was like creating a whole new skill.”

Rivian said the chips will be manufactured by Taiwan Semiconductor Manufacturing Company.

Rivian's R1 “ground truth” vehicle, equipped with additional sensors and lidar, could be seen in the parking lot of the company's Palo Alto campus.

This in-house vehicle collects data and essentially measures the real-world environment, from roads to objects on the road. Engineers can use that data to evaluate the accuracy of Rivian's self-driving system.

Rivian plans to release an R2 model equipped with lidar in the second half of 2026. The total sensor stack includes 11 cameras, five radars, and one LIDAR, a sensor that uses laser light to measure depth.

Philbin said every lidar-equipped R2 sold will be a “ground research vehicle,” providing continuous training data for the company's self-driving systems.

Unlike the rotating sensor on top of Waymo's robotaxis, LiDAR is solid-state. Rotation sensors have many moving parts and can be relatively expensive.

Rivian engineers told reporters that the company kept aerodynamics in mind when integrating all the sensors to reduce excessive soil accumulation. Engineers say there are also heating elements and soil detection systems to consider for owners in colder regions.

Philbin told Business Insider that adding LIDAR was a “no-brainer” decision. He said the sensors will significantly reduce cost and safety is critical for autonomous driving.

Rivian developed Rivian Assistant to allow drivers to retrieve information from various Frontier LLMs at their request, Bensaid told Business Insider.

For example, a command to find the nearest coffee shop depends on a different model than a command to lower the temperature inside the car.

Pranil Vora, Rivian's senior manager of AI platforms, demonstrated to reporters that the assistant can understand commands said in natural language without the user having to refer to specific buttons or functions in the car.

For example, when Vola asked, “It's too hot. Can you make it a little cooler for everyone on the plane?” Rivian Assistant lowered the cabin temperature.