Overview of Decoder with The Verge: “Rivian's software chief thinks you don't need CarPlay or buttons”
In this episode of Decoder, Neil Patel talks with Wasim Ben-Sayed, Rivian’s Chief Software Officer and co-CEO of Rivian Volkswagen Group Technology (RV Tech), about how Rivian is building the software stack for next-generation EVs and why it believes the future of in-car computing is voice- and AI-first rather than button- or CarPlay-first. The conversation centers on Rivian’s software architecture, the Volkswagen joint venture, the new Rivian Assistant, and the company’s long-term bet that integrated, agentic software will make traditional phone mirroring and physical controls less central to the driving experience.
Key Topics Discussed
Rivian + Volkswagen joint venture (RV Tech)
- RV Tech was created to bring Rivian’s software, electrical architecture, and engineering culture into Volkswagen Group vehicles.
- The JV is meant to scale Rivian’s technology across a huge portfolio, from mass-market vehicles to premium/luxury brands like Audi, Porsche, Bentley, and Lamborghini.
- Wasim says the JV is responsible for the underlying electrical architecture and operating system, while individual brands get customization layers on top.
What “software-defined vehicle” means
- Wasim defines a software-defined vehicle as one where the car is designed around:
- centralized, powerful zonal computers
- fewer ECUs
- software-first development from the start
- He argues the old model—many suppliers, many isolated systems, waterfall development—can’t keep up with modern expectations.
- Rivian’s approach is to build the car around the software stack, not bolt software onto a traditionally designed vehicle.
Why the VW partnership can succeed where past efforts struggled
- Wasim says Volkswagen recognized it needed a clean-sheet approach rather than incremental fixes.
- He emphasizes two requirements:
- real software talent that can build an operating system, not just abstracted functions
- a cultural shift toward agility, fast iteration, and cross-functional ownership
- He frames the JV as successful because VW leadership reportedly committed to preserving “the Rivian way” rather than forcing Rivian into a legacy model.
How RV Tech is organized
- RV Tech has about 1,500 employees.
- The team started with roughly:
- 800–900 people from Rivian
- about 150 from Volkswagen
- the rest hired over the last 18 months
- Everyone works for RV Tech, not directly for Rivian or VW, which helps create a shared operating model.
Rivian Assistant and the AI strategy
- Rivian’s new Assistant is designed as a deeply integrated vehicle layer, not just a chatbot on top of the UI.
- It can control many vehicle functions, including:
- drive modes
- ride height
- navigation-related actions
- some climate and convenience features
- Safety-related and regulated functions are intentionally blocked or restricted.
- Wasim says Rivian built its own orchestration layer so it can control which functions the assistant can access and maintain safety/reliability guardrails.
Voice as the future primary interface
- Wasim doubles down on his earlier view that buttons are not the primary future interface in cars.
- His argument:
- driving is a hands-busy, eyes-on-road environment
- voice is a more natural interface when it works well
- modern foundation models finally make conversational interaction viable
- He sees buttons as secondary, not obsolete, with some tactile controls still useful for specific tasks.
CarPlay and Android Auto skepticism
- Rivian remains committed to not offering CarPlay as a general screen-mirroring solution.
- Wasim argues CarPlay takes over the full screen and prevents Rivian from delivering a more integrated, end-to-end experience.
- He points to falling customer demand for CarPlay in Rivian surveys as the company’s own software improves.
- He believes agentic AI will eventually make the CarPlay debate less relevant.
Agentic integrations and the Google partnership
- Rivian Assistant uses an MCP-based integration with Google Calendar.
- Wasim describes the assistant as an agent orchestrator that can share context across apps and services.
- He imagines future use cases like:
- planning a route with charging stops and restaurant preferences
- adding a trip summary to your calendar
- texting details to a spouse
- preconditioning the car based on schedule data
- Rivian currently uses:
- internal models for local/edge functions
- Google models, including Gemini, for cloud-powered capabilities and grounding
Edge AI, connectivity, and the R2
- Current Rivian R1 vehicles rely heavily on cloud connectivity; local interactions are more limited.
- The upcoming R2 will bring:
- 5G
- much more powerful local compute
- up to 200 sparse TOPS for AI
- Wasim says this should reduce latency, improve reliability, and make more assistant tasks happen locally instead of in the cloud.
Cost, compute, and tradeoffs
- Wasim says Rivian treats AI compute as part of a “differentiation budget” alongside exterior, interior, and other vehicle priorities.
- He argues that more local compute can ultimately reduce cloud inference and connectivity costs.
- He also says the architecture is designed to be model-agnostic and adaptable as AI providers change.
R3X and product roadmap
- When asked about the R3X, Wasim says it’s “here” conceptually, but the practical priority is still the R2, which is the more important near-term volume product.
- He frames the R2 as a necessary step before Rivian can bring more emotional/niche products like the R3X to market at scale.
Main Takeaways
- Rivian is betting on a vertically integrated software stack rather than relying on CarPlay-style phone projection or third-party interfaces.
- RV Tech is a major strategic expansion of Rivian’s architecture and culture into Volkswagen Group.
- AI is becoming the centerpiece of Rivian’s in-car experience, with local edge compute, cloud models, and agentic integrations working together.
- Safety and reliability still set hard limits on what the assistant can do.
- Voice is the long-term interface bet, though some physical controls will remain for practical reasons.
Notable Ideas
- “Software-defined vehicle” is less about a buzzword and more about moving the car from a supplier-assembled machine to a centralized computing platform.
- Rivian’s assistant is meant to be the connective tissue between the car, the cloud, and a user’s broader digital life.
- The future, in Wasim’s view, is not one app taking over the whole screen, but many agents working together through a shared orchestration layer.
- He believes the biggest value of the VW partnership is not just licensing technology, but exporting Rivian’s process and culture at scale.
