Rivian's software chief thinks you don't need CarPlay or buttons

26:47Nilay Patel:We need to take a quick break. We'll be right back.

26:58Nilay Patel:support for this show comes from shopify starting something new isn't just difficult sometimes it can be truly intimidating you're putting in all of this time and energy without knowing for sure if it's going to pay off but there's always another way to look at it what if it does work what if your gut has been right the whole time that's where shopify comes in Shopify powers millions of businesses around the world and accounts for nearly 10 % of all e-commerce in the U.S. Everything from established names like Allbirds and Heinz to brand new businesses just starting out. Their design tools make it easy to build exactly the kind of online presence you're imagining.

27:37Nilay Patel:With hundreds of templates ready to go. And with built-in marketing features, you can spin up email and social campaigns in just a few clicks. So you're reaching customers where they already are. Start your business today with the industry's best business partner, Shopify, and start hearing. Sign up for your$1 per month trial today at shopify.com slash decoder. Go to shopify.com slash decoder. That's shopify.com slash decoder.

28:12Nilay Patel:Support for this show comes from Adobe Acrobat. We all know sending a file is easy. Making sure your clients understand the file is the hard part. But with PDF Spaces and Adobe Acrobat, you can give your clients the full picture with custom intros, audio summaries, and a helpful AI assistant. So if you want to stop the endless follow-ups, do that with Acrobat. Need to make your dots crystal clear? Do that with Acrobat. Want to make sure your clients get everything they need to hear? Do that with Acrobat. Learn more at adobe.com slash do that with Acrobat. Support for this show comes from Doppel.

28:53Nilay Patel:Maybe that ping you just got is an urgent message from your CEO. Or maybe it's a deepfake trying to target your business. Doppel is the AI-native social engineering defense platform fighting back against impersonation and manipulation. As attackers turn to AI to power increasingly sophisticated strikes, Doppel uses it to fight back. Their digital risk management dismantles attacker infrastructure, while human risk management builds team resilience through simulation and training. With automated takedowns, multi-channel coverage, and AI defenses that build intelligence with every fight, Doppel works relentlessly to protect people, brands, and trust.

29:31Nilay Patel:Doppel, outpacing what's next in social engineering. Learn more at doppel.com. That's D-O-P-P-E-L dot com.

29:49Nilay Patel:we're back with rivian software chief wasim bensayed before the break we were discussing his role as co-ceo of rivian's joint venture with wokswagon and how that organization is structured now i want to ask the other decoder question how does wasim make decisions let me ask you the other decoder question and i want to turn to the software itself uh i ask everybody this question we've talked about a little bit how do you make decisions what's your framework making decisions? I think the number one rule, I mean, obviously, my job every day is making decisions. But there's a number of guidelines that I try to apply.

30:23First of all, in terms of coaching with my team, I tried to push decisions as much as possible to the lowest levels of the organization. And I think really one of the anti patterns that I see in a bunch of companies where they try to bubble up decisions to the highest levels of executives. And that tends to create a culture where things are slow and then employees don't feel really empowered. Now, in cases where I personally have to make the decision, few guidelines to the team. Never come to me with one option. Show that you went through an analysis, have multiple options, and then make a recommendation.

31:03I want a culture where I empower my team to be really a force of proposal and then come up themselves with recommendations. Now, the rule that I use in terms of how much time should I spend on a decision is really, is it a one-way door decision or is it a two-way door decision? If it's a two-way door decision, then I don't need to spend that much time on it. It doesn't need to be really a hard framework. We don't need to go into an extreme in some cases of collecting tons and tons of data so that we get to a decision. And in some cases, I really just use my gut. and I'm a product builder at heart and I know that some of the decisions, even if the data are against me, then I would go with my feeling.

31:58And in some cases I'm wrong and I'm the first to recognize that. Now, if it's a one-way door decision, then that's a different process which requires much more preparation, requires much more data and then arbitration for how we do things.

32:13Nilay Patel:Give me an example in this context of a one-way door decision at RV Tech and a two-way door decision.

32:24there's multiple of them. And I think probably one of them is like leading to probably the next topic of discussion that we will have, which is our overall approach around AI. And we had a ton of debates internally of whether we should basically just use a third-party AI solution or develop our own solution. And there's a ton of tension because you look at the investments in the AI world and you would think that with everything which is happening, this is obviously a hard problem to solve. Now, it was really clear for me, given the opportunity, given how transformative this is for the entire user experience, that we really need to own our destiny in terms of having a platform that allows us choice, that allows us to change foundation models as we wish and own that integration layer that allows us to power the entire car operating system.

33:41Nilay Patel:So this is Rivian Assistant. I've been playing with it for a few days now. What do you think? Truly had some searching conversations with it. just to push the boundary of what it can do. It's super interesting to have a car where, you know, even in the interface, it does the wavy line on the main screen. It's very much the car is running its assistant, not an overlay. You can tell. The assistant can go and address lots of parts of the car. And then there are places where it can't or it won't. And actually, I think one of the most interesting things about it is that it won't tell you why it can't do things.

34:14Nilay Patel:It is insistent that it won't tell you why it can't do things. Don't worry. I have very specific questions about that. But it strikes me that this is a natural evolution of, okay, the whole car is run by a finite set of computers. And that means our assistant can just run around and talk to those computers and talk to the functions of those computers control in a way that, I don't know, if you try to glue an assistant onto, I have a Cadillac EV. If you try to glue an assistant onto that, it has to like go talk to its CCUs. Like it's just very obvious that something else is happening with Google Assistant in that car.

34:48Nilay Patel:That's the opportunity. The assistant can talk to the whole car. Then there are places where it just can't for some reason. I'll give you one example. It just struck me as very odd. I was driving in the rain and I said, hey, turn on the back window wiper. And it just won't. And I thought, is that a safety reason? Is that you just don't know how to do it? You're like lost in the zonal architecture. And I asked it and said, I can't tell you why I can't do these things. But here's where the button is, which is really interesting for an assistant at car to do. It's like, I'm not going to do it for you, but the button is on the stock.

35:20Nilay Patel:Push the button. How do you make those decisions in the context of an assistant of what it can and cannot do? Awesome. So there's a lot of things here. I mean, first of all, I think you described it really, really well. Our approach, our philosophy for the Rivian Assistant is not just put a chatbot and then slap it on top of the UI. It's really developing what will become the connective tissue that enables our users to interact with pretty much every single feature in the car. And even more than that, bring their own personal digital ecosystem in the car through the authentic integration. Now, to your question, what it can do and then what it cannot do, us controlling the wiper, it's obviously possible.

36:09And I'm sure that you have seen that it can do way more. It can change your drive modes. It can change your ride height. Right.

36:16Nilay Patel:I could raise and lower the car at 55 miles an hour with the air suspension, which was cool. and it was like the slowest low rider experience you could possibly have. And then I couldn't turn on the wiper. So what is the split? Yeah. And that's honestly, that's one of my favorite features. And the way I like to interact with it is I don't tell it to change the right height. I tell it, okay, give me drive mode with more pep. And then it does it and it changes to sport mode. I mean, this is really the magic of having that true conversational experience. Now, the reason it does not do wiper is by design, we actually block a number of features which are safety related.

36:56And cars are obviously homulated, regulated. So everything related to wipers, things related to windshield controls, things related to highway assistance. Those are regulated functions, which today through our framework, we block them for safety reasons. And safety is really one of the core tenets in how we develop the entire experience.

37:20Nilay Patel:So the other one that struck me, you know, cars have rear seat sensors. We have kids. So every time I get out of my car, it reminds me there might be a kid in the backseat because it has sensed the weight in the backseat. I think this is one of the funniest sensors any car can have because the car seat is always in the backseat. So it's always reminding me that the kid might be in the car. So I asked it, is anyone in the backseat? and it maybe this is just a bug but it said i i'll find out and then it said i can't access that sensor and i said what sensor are you trying to access and it refused and i probably had a five-minute argument with your assistant about why it wouldn't tell me what sensor it was trying to access the reason i'm asking this is not because it's a bug or i really needed to know if anyone had car seat at the time i'm just very curious as you think about building the the assistant that can access all of the sensors and the architecture of how that might work and how we might interact with cars.

38:15Nilay Patel:There's just a moment where maybe it's for safety reasons, or maybe it just can't, doesn't work right now in the version I have, where the LLM has to go talk to another computer and that computer has to give it permission. And I don't know if anyone in any part of the tech industry has figured out exactly how that should work. And I'm just wondering what your point of view is. Yeah. And I think in this specific case, it should have actually told you what's in the back seat. So that's a bug. That's on me. No, it was like, I'm not telling you what sensor I'm trying to get to. And I was like, why?

38:44Nilay Patel:And it was like, I just, I'm not going to tell you what sensor I'm trying to get to. Yeah. That one is on me. And I think the beauty here is that, I mean, we have it even house and then we'll be able to calibrate that answer and then we'll fix the bumper and send you a note here next time. That's very good. Every time we get a car executive on the show, I just complain about the experiences I have. It's perfect but that but that piece where it's how do you talk to the computer and what permission does that other computer give you every assistant at every level is running into that specific barrier and i'm just curious what you think the answer is yeah i mean pretty much i think think about our architecture um this way the assistant has deep integration into the entire vehicle operating system so in theory uh i mean with the integrations that we have built unless we have a bug like the one that you have, you should be able to do everything.

39:35The only functions which are not allowed are functions which are safety related because obviously of the homologation reasons, but then also functions where we think that the level of reliability that we can get from the LLNs does not make us comfortable to expose them to the end users. But that's really like the beauty of the internal in-house orchestration layer that we have built, where we have a ton of guardrails, which allow us to control which functions are exposed by the assistant or not.

40:12Nilay Patel:All right. You mentioned that I was going to get you in trouble. I'm going to get you in trouble again. In 2024, you said using buttons in a car is an anomaly of modern design. And people love buttons in their cars, so you got in trouble for saying that. But the thing you said was that voice should be the future. Okay, this is the first gesture at voice being the future. Is it good enough? Because we're right on the cusp of, are these things actually good enough to build the kinds of products people want? I think we are on the cusp of something really big. And I deeply believe, like when you think about it, you're in the car, you're driving, you're focused on the road.

40:48So in theory, in theory, the primary interface with which you should be interacting with the car is actually your voice. The only reason that drivers and consumers so far did not interact with the car through voice is that, like to put it really bluntly, the technology has been broken. And I think that's really the beauty of what we have now through the technology disruption, which is coming with the foundational models. the foundational models are providing us this wonderful opportunity so that we can truly get to a conversational experience and where drivers can interact with the car in human language.

41:33I don't need to tell the car, open the front. I can say, open the front front, or actually I can say, I have a bag in front of the car and it will actually open the front. And I think that completely changes the way you interact with the car. On top of that, we have now the opportunity with all the agentic framework so that we truly can give people their time back in the car. And I hope you tried our Google Calendar agentic integration. But you can imagine how the experience will be in the future where you're driving and then you should be able to perform operations on your calendar. You should be able to perform operations on your email.

42:24And then in the future, with the framework that we have with the agent to agent integration, you can actually interact with many more apps from your own digital personal ecosystem.

42:36Nilay Patel:Can I ask you about the word agentic in this context? So the way the Google Calendar integration works with Rivian Assistant, just describe it quickly for people. It shows you a QR code. You connect your Google Calendar to it. And then Rivian Assistant can read your calendar, add events, remove events, do calendar stuff. I'm very curious how that's agentic and how it's built such that word agentic is meaningful. Because I have like 500 apps over the past 10 years that can do Google Calendar stuff through the standard API. So one, how is it agentic? Is it MCP? Is it something else? And why build it that way versus just doing a bunch of API integrations?

43:13I mean, you can actually build it with an API integration. I think the advantage of an agency integration is that you can share the context, and then you can perform multiple integrations within the car. And in this case, it's based on, like, this is an MCP integration, which also allows us, I mean, think about it in a way where you can imagine that in the future, instead of having that mono access to every single app on your car, or honestly, to that extent, even on your smartphone, you can start aggregating and connecting many of those apps through the agentic framework and have them present a unified user experience.

44:01and this is how we're able to connect the navigation to calendar, for example. Like I can go to the assistant now and I can tell the assistant, okay, I want to plan a trip from San Francisco to San Diego and I want to have two charging stops. I want them to be close to an Italian restaurant. I love Italian food. The assistant would go and play that and then I'll say, okay, okay, print the summary and add it to my calendar and then send it as a text to my wife. And this type of integrations, when you have behind the scenes, the agentic framework can really allow many more capabilities, which in the future as well, and this is where agentic can be even further utilized.

44:58you can start going into more autonomous functions in the sense that, let's say, you have an invitation in the calendar with XYZ details, and then you can start having reminders saying, okay, do you want to go to this place? Or you're actually really late to your meeting. Do you want me to start preconditioning your car? So that's the beauty of bringing the depth of the Agentec integration.

45:29Nilay Patel:We need to take on a short break. We'll be back in a minute.

45:40Nilay Patel:Support for the show comes from Upwork. Scaling a business takes the right expertise at the right time. Upwork helps growing teams quickly bring in specialized freelancers so you can move faster and take the business to the next level. Upwork is a one-stop platform to find, hire, and pay expert freelancers, letting you delegate and then just keep it moving. You can find specialized talent across web and software development, data and analytics, marketing, business operations, and more. Browse profiles, review their past work, and get help scoping the role so you can hire with confidence. You can also enjoy the benefits of Business Plus, which gives you access to the top 1 % of talent on their platform.

46:23Nilay Patel:With AI Power shirtlisting, you'll get matched to the right freelancer in under six hours. Skip the endless searching. It's free to sign up and posting a job is easy. Visit Upwork.com right now and post your job for free. That's Upwork.com to connect with top talent ready to help your business grow. That's U-P-W-O-R-K.com. Upwork.com.

46:50Nilay Patel:Support for the show comes from ServiceNow. AI was supposed to handle the parts of the job you hate. Instead, it just describes them, suggests what to do about them, and then leaves you to do it. That's not help. That's homework. ServiceNow's AI specialists are different. They're not a tool. Think of them as digital teammates who actually do the work from start to finish. Cases get resolved, requests get processed, loops get closed, and most importantly, no extra work for you. because when you can truly delegate to AI, you can get back to the work only you can do. The work that requires a person with ideas and judgment, and you know, a pulse.

47:32Nilay Patel:To learn how to put AI to work for people, visit servicenow.com.

47:40Nilay Patel:Support for this show comes from Vanta. If you're a business owner, you've likely noticed more customers asking for proof of security right from the start. It's a sign that risk and regulation are only increasing, and earning that customer trust is essential to closing deals. But that can be difficult, expensive, and a major drain on your time. With Vanta, you can automate compliance by unifying risk, security, and trust in a single AI-powered platform. It takes the manual work out of your security and compliance process and replaces it with continuous automation. So whether you're prepping for a SOC 2 or running an enterprise GRC program, Vanta keeps you secure and keeps your deals moving.

48:20Nilay Patel:Vanta says that companies like Ramp and Writer spend 82 % less time on audits with Vanta. That's not just faster compliance. It's more time for your business to grow. You can get started at Vanta.com slash decoder. That's V-A-N-T-A dot com slash decoder. Banta.com slash decoder.

48:51Nilay Patel:We're back with Rivian Chief Software Officer with Sim Ben Sayed, discussing the ins and outs of the company's new AI-powered voice assistant. So I think I understand that, right? You're in your car. It can go access a bunch of apps and services you have. You can take actions across them. You've collected a lot of data in one place. This brings me to like a very deep sort of existential question whenever anyone talks about ambient computing in this way. Where does the logic live, right? The idea that you're going to have that interaction in your car and not at your laptop or on your phone, that seems like a big jump to me, right?

49:26Nilay Patel:The same way that when the smart speaker companies would be like, you're going to talk to your thermostat. And I'm like, why? I'm going to talk to my phone. I don't feel the need to talk to my thermostat in this way. Do you think people are going to do that in the car? Are you going to bring Assistant to the Rivian app on a phone? Can you compete with the series and Google Assistants in that way? How is that all going to work? I think, actually, like the way I think about it, I think it would be both. And I think this is really the big difference between the old world where we had unique applications and then the new world where we have agentic integrations.

49:59And think about the Rivian assistant as an agent orchestrator, which has privileges because it can deeply integrate with the vehicle controls, with the vehicle operating system. It understands safety, it understands which things to do, which things not to do. And nobody else can develop that better than us because we develop the entire vehicle software. But then at the same time, it has interfaces and connections to other agents. And this is just the beginning. I think, think about it in a way where in the future, you can actually probably be able to bring your own personal favorite assistant and chatbot to the car.

50:47And then it can share context with the Rivian assistant. I mean, this is really the possibilities that this new world, this new type of integrations is allowing us to do.

51:02Nilay Patel:Is the Rivian Assistant the kind of thing that is possible because of the RV tech software stack? And is it possible that we'll see the Rivian Assistant or something just like it in Volkswagen's as well? Or is this special to Rivian? This is special to Rivian. Everything which is AI stack is developed uniquely for Rivian. I mean, this is Rivian's brand priority as we see cars becoming more and more AI-defined. But we're in discussion so that we can have similar technologies for the Volkswagen group. Rivians are sort of famously on LTE. When I first got in this car, I saw the LTE indicator, and I thought, oh, something wants to be wrong.

51:45Nilay Patel:And I drove around, and then I realized, oh, it's just LTE. Is there any latency concern with that, especially with voice and going out in the world and doing whatever inference you need to do? So two things. One, we need to get you an R2. The R2 has 5G. There you go. It's coming soon, and it's amazing. And then two, and I think your touch base really on one of the architecture considerations for the technology, which is when you look at vehicles like the Rivian R1 today, most of the interactions will happen with the cloud. So as you say, it's connectivity dependent and then they will happen best when there's a strong connectivity with the external world.

52:33Now there's a number of interactions that happen locally with the car. And if you tell the car, I am cold, that interaction is being managed by a small language model that sits directly on the edge. Now, the beauty of what will happen next as we get to R2, not only we will have 5G, but then edge AI will be way more powerful and capable on the R2.

52:59Nilay Patel:Just to be clear for the audience, when you say edge, you mean local, right? It's running locally in the car. So meaning that the local computer on the R2 will have up to 200 sparse tops of compute dedicated to AI. I know this sounds like extremely technical, but think about it as this is more capable than some of the self-driving platforms today. And it's more capable than the amount of AI compute that you have in your smartphone. All of that will be available locally in the R2 car, which is coming soon. And that allows, as you mentioned, to basically not face such limitations and issues with connectivity, but truly get to much lower latencies because a lot of the processing will happen directly on the embedded locally and get to a truly conversational experience where it's pretty much instant.

54:02Can I just ask you a very in-the-weeds question?

54:05Nilay Patel:We're talking about putting compute in the car. We're going to do some amount of local inference in the car. GPUs are expensive. RAM is expensive. How much of the bill of materials is RJ giving you to do all this in the car versus, I don't know, bigger motors, bigger batteries? How much of the range can you pull off to do local inference in the car? I mean, this is kind of the trade-off you're talking about. Yeah, I mean, this is what I love about RJ, which is what has always attracted me with RJ is that RJ thinks big things long term. And he knows that in this case that the world is moving to AI.

54:42And this is why, like the way we do decisions, for example, from a material standpoint, it's a very hard process where there's a ton of trade-offs all the time. And you can imagine the tension between people wanting to push for a better exterior part and then people wanting to push for a better interior part and then people wanting to have better technology. And then we have what we call the fun or the differentiation budget in the car. And RJ, absolutely, it was no discussion that we will equip the car with higher inference compute and then higher memory because this is really the future. And it's an opportunity for us to completely reshape the way people interact with their cars.

55:31And actually, in the long run, to be honest, it solves itself from a unit economic standpoint, because as we do more and more interactions locally in the car, we avoid the back and forth with the clouds. We avoid the connectivity costs. And then we also don't have to pay for the inference costs on the cloud. So in the long run, it's actually economically positive as well.

55:57Nilay Patel:That's not just a spreadsheet you made up to win an argument that actually models out? Kind of. Because I could probably make up a spreadsheet to win an argument. The reason I asked it, because that was my next question, was about inference costs, right? They're going up. There's rate limits at all the big providers. What model are you using right now? What are the frontier models you're using right now? Actually, like the architecture that we have built is not model dependent. So really, one of the foundations of the architecture is it allows us to interact and plug and play pretty much with different foundational models.

56:28and similarly it can use different modalities in the way users can input their requests, whether it's voice in this case, but then it can use vision as well. It can use text if we want to enable that. When it comes to the models themselves, we use currently a combination of internal models for everything that runs locally on the edge And then we use models from Google. We have a partnership with Google. Things are going really well in terms of deep access to the advanced Gemini models, as well as the grounding of results, which is also powered by Google.

57:18Nilay Patel:This is another question. I asked it to just, what are the top five headlines in The Verge? Just to see, can this thing browse the web? and it returned some results. They were just, I think they're 24 or 48 hours old, right? It was our top five headlines from yesterday. Does this thing have like a web browser in the background or is it just pulling from a Google data corpus? How does that work? It actually can, it should in theory connect real time. And then this is where the grounding with the Google results come into the picture. And in theory, it should give you the latest headlines. So if it didn't, then that's another one on me.

57:54Nilay Patel:Well, I was just curious, It's like there's some, you know, lots of people have this experience now where like the data in the model is old. And then there's some cutoff. And I was just trying to find the cutoff. And then I had a long searching conversation with it about we need to buy a new air conditioner. And I was just asking it to do math about air conditioner efficiency. This is very boring, but this is what I talked to your car about for a while. And it just occurred to me, like, I am making it think very hard. Like, I'm burning, like, I am wasting more energy talking about how efficient of an air conditioner to buy.

58:24Nilay Patel:Like, this is not a good ratio of energy spent to save energy over here. How does that work? I mean, you have to pay a monthly fee for the connectivity package to access assistant, but then I might burn way more tokens than that fee could ever pay for. How does that math work out? It really depends. It's a great question. Now, obviously, in these cases, there's all sorts of techniques of what we call rate limiting that we can apply. Like if we have seen in your case that you're spending 20 hours discussing with the assistant, then we may do something behind the scenes. Similarly, in the way we configure the models, we, I mean, given the types of interactions that you have in the car, you would not be interacting with the latest and greatest, say, Opus 4.7 models so that you bring, you burn so many tokens.

59:17And then a lot of it also depends on the aggregation across users in terms of the types of requests, as well as arbitration that we do between the edge and the cloud. And as I mentioned, the more we move to edge and local compute, the better it is for us in terms of overall inference costs.

59:39Nilay Patel:So let me just ask you this question again. Now that you've shipped the software, people are using it. You're getting extremely detailed feedback from me. Do you still think having buttons in the car is an anomaly?

59:52I think, I deeply think that voice has the chance to be the primary interface in the car. I also think that buttons can exist, but then they shouldn't be the primary way with which you interact with the car. I think there's much more that is possible through voice where you can not only do one single function. You don't have to go and fiddle with so many functions. You don't have to go even deep into the touch screen to look into specific features. A great voice experience can really elevate all of that and allow users to talk to the car as a human would do and really take the overall experience to the next level

1:00:48Nilay Patel:are we going to get the h-track buttons back in any future rivians that's really what i'm asking here actually with the r2 we have a great way to have a tactile feedback uh for a big paddles on the wheel yeah yeah they're really awesome uh it's a good pivot but i'm asking are we going to get the fan speed button back in the center stack. In the center stack, no, but we have the same in the Halo wins. It's a great compromise. You knew it was coming. I have to ask you about CarPlay here because it strikes me just as you imagine this future of the car where it's connected to your calendar. It's connected to all this context.

1:01:25Nilay Patel:It has autonomy, which is something you're also working on. You get in the car and it knows it's time to go to work and you just say, let's go. And the car takes off driving. This is when you would use like a vast number of applications, right? This might be when you have the focus to push the buttons again. I'll just make that argument. But this is when you would want a whole number of apps. And I hear from our readers every time I talk to a car executive, the reason I want CarPlay is because there's 5 ,000 apps on my phone and no car OEM is ever going to support them on the built-in infotainment.

1:01:56Nilay Patel:This is where you would say, okay, project your phone to the center stack. The car is driving itself. Have at it. Phone projection all day. Do you think that the tide is turning? Are you still absolutely committed to not having CarPlay and Rivian vehicles? I think, first of all, it's really important to go through our philosophy of how we see software in the car and how we see user interface. The challenge with screen mirroring solutions is that they take over every single pixel in the car. And that's not the way we see ourselves interact with our users. You drove our car four years ago, and you drove the car in the past few days.

1:02:41And I hope you have seen how much change happened in the car. And that's truly by bringing the end-to-end features. I mean, not only changing the user interface, but having your navigation know exactly about your drive mode, know exactly about your efficiency offering that level of convenience is really what is resonating with a lot of our customers. I mean, even if I look at our own internal statistics, five years ago when we first shipped the R1T and the R1S, the number one request from customers was CarPlay. And at the time, like we do all sorts of survey with customers, more than 70 % of customers were requesting CarPlay.

1:03:26In the recent survey, that number is less than 25 % now because with the level of features that we have shipped to customers, with the level of end-to-end integration, with the level of convenience that we are bringing, CarPlay is, or Android Auto to that extent, is no longer the topic of discussion. And I think what we're actually seeing right now with the advancement of AI technologies is just another reason why I deeply believe that RJ and Rivian, we made the right choice in terms of investing into our own technology and our software. Cars are moving from, as you said, the buzzy world of software defined to AI defined.

1:04:22And the possibilities that we have now to have such deep AI integration into the car, it will make the entire car play debate in the future completely obsolete. I really believe that the way you interact with apps, which are monothreaded, single buttons, single icons, a lot of that will be now completely reshaped into a world where it will become an agentic integration that presents itself into a wholesome user experience to the user.

1:05:00Nilay Patel:I buy that in the big picture, but give me an example of that. I'll just put up an idea that I get from our readers all the time for you to react to. There are tons of little apps. They're basically media-playing apps on phones. And it's trivial to push the button and do the CarPlay app. The one that I always think about, we got an email from a reader, and he said, I have a Bible app that is never going to be inbuilt into anyone's infotainment system. It's made by a small developer, and I love it, and that's why I need CarPlay. And I'm always going to buy a car with CarPlay because of it. That is about as small of an edge case as you get.

1:05:31Nilay Patel:but this one customer is going to pick a car based on it. Are you going to make that developer build an agentic AI integration into the Rivian Assistant, or are you just going to lose that customer to CarPlay? I mean, this is the beauty of the technology disruption in which we are today. The answer in that case does not necessarily need to be we will build an agentic integration for that particular app. It can be. If it is like, say, a Spotify or an Apple Music, absolutely. But then if it's a small app, the answer to that could be that from the Rivian assistant, we have an integration to your favorite voice assistant on the car.

1:06:12And then you can ask the voice assistant on the car to play that particular app through Bluetooth audio. That is possible as we open up the framework and as we allow more integrations to bring your own digital ecosystem to the car.

1:06:26Nilay Patel:So you're saying I've got, we'll use Google because Gemini is more present on an Android phone than Siri is currently on an iPhone. So you use Google and they're also your partner. So you're saying you talk to Rivian Assistant, Rivian Assistant knows about your Google account and your Android phone, and it's going to go talk to Gemini. And Gemini is going to go operate your phone and stream Bluetooth audio to the Rivian Assistant. In the future, all of that is possible. Is that better or worse than phone projection, right? This is like a different kind of loop than just saying, put the interface here and let the user do it.

1:06:58It could be possible through phone projection. I think the challenge with phone projection, again, is that, I mean, first of all, you have to go, like, as you're driving, you have to go through your phone. In some cases, you have to press, like, multiple buttons so that you get to the menu of your own app. The other thing is that it takes over the entire screen. and that is a degradation of the experience. Now, is the alternative solution available right now? No, but I think the beauty of the wave of technology in which we are is that we finally have the building blocks so that we can really redefine those type of interactions and we can allow hooks now into your own personal device through a different way of interaction rather than truly integrating end-to-end the app on the car itself or take over the entire screen from your phone.

1:07:58There's a third path now, which is possible.

1:08:01Nilay Patel:I mean, obviously, I think it might be easier with Google. Again, they're your partner. But where would that connection to Google Assistant happen? Would I be like, I don't think I'm holding up my Android phone to the speaker and letting the assistants talk to each other out loud. Although that would be fun. I mean, that would be deeply hilarious to hear the two assistants just have a conversation. Like, can you please play the music app for me? Does that happen in the cloud? Does it happen locally? Where does that integration point between assistants happen? I mean, think about it as the assistant in the car will know how to talk to your, say, Gemini, your own personal assistant.

1:08:38And then in that case, your own personal assistant would be controlling your phone.

1:08:45Nilay Patel:The reason I'm asking this in this way is at some point you have one main assistant and all the other things are sort of like agents that can talk to. And then maybe no one's ever talking to Ravine Assistant again, right? Like this is kind of the, you pull that thread all the way and Gemini is just doing everything for you all the time. Is that a danger or are we just nowhere close to even having to worry about that? Honestly, we don't worry about that because we know the opportunity that we have and we know the level, the breadth of capabilities that we can offer. No other assistant will be able to know as much as there will be an assistant about the car controls.

1:09:23None. Similarly, the fact that we have the surface of integration sitting in our own operating system enables a ton of opportunities in terms of use cases that you cannot simply do with your phone or you cannot simply do by calling another assistant. Like imagine, okay, I'm driving and then in the near future, we should be able with the technology to have an agentic integration with your favorite food delivery. And then you would say, okay, I need to be, the car knows exactly exactly when you will be home, you'll say, okay, order my favorite sandwich from XYZ shop, your account is already configured, and then the assistant will pick the destination, will get you to your favorite restaurant.

1:10:25All of that is integrated. You just need to do it through one voice command. Like those type of experiences where things become so seamless, where things become so easy, as if you're talking to a human, which connects the dots across multiple surfaces of your digital ecosystem, would only be possible through such integrations. All right.

1:10:50Nilay Patel:Listen, we're out of time here. As you can tell, I can obviously talk to you about this forever. I don't think anyone has figured out how all this works, and it seems like you're making some big decisions. So you're going to have to come back as you've learned how this goes after this is shipped to all of your customers and certainly when the R2 is out. There is one question that I have to ask every single Rivian person that I encounter. It is very important to me. When is the R3X coming out? It's here. Do you see it? When can I get one? By the way, it's my favorite car. And like I keep asking RJ that question all the time.

1:11:28Now, you talked about decisions, you talked about trade-offs and us delivering the R2 before the R3X, you can imagine it's a big decision. It's also a hard decision because in our hearts, we all deeply want to have the R3X as soon as possible. Now, we also know that the R2 has the best ingredients to be a wildly successful car. And the U.S. needs another alternative for a great SUV for families. And this is what the R2 brings. And as we ship the R2, as we scale our volume as a company, we will earn the ride so that we bring fantastic and such emotional cars like the R3X.

1:12:18Nilay Patel:I know that that is in one way the right answer. I'm just saying for me personally, come on. Just send me one. It'll be great. You can do it. I'll give you more feedback just like this. I will break your R3X prototype in 10 ,000 different ways. You'll get the bug reports. It'll be great. Tell RJ I made the offer. Awesome. I'll get one at the same time as you. Sounds good. Well, Sim, thank you so much for being on Decoder. This is great. Thank you. I'd like to thank you for Sam for taking the time to join Decoder and thank you for listening I hope you enjoyed it if you'd like to let us know what you thought about this episode or really anything else at all drop us a line you can email us at decoder at the verge.com we really do read all the emails or hit me up directly on threads or blue sky we're also on youtube you can watch full episodes at decoder pod and you can watch clips on tiktok and instagram also at decoder pod that's a lot of fun if you like decoder please share it with your friends and subscribe over your podcast decoder is a production of the verge and part of the doxing podcast network The show is produced by Kate Cox and Nick Stat.

1:13:16Nilay Patel:It's edited by Xander Adams. Our editorial director is Kevin McShane. The decoder of music is by Brickmaster Cylinder. We'll see you next time. Did you know if your windows are bare, indoor temperatures can go up 20 degrees. Turn the temperature down with Blinds.com and get up to 50 % off custom window treatments like solar roller shades and more during the Memorial Day mega sale. Whether you want to DIY it or have a pro handle everything, we've got you. Free samples, real design experts, and zero pressure. Just help when you need it. This is your last chance to shop up to 50 % off site-wide during the Memorial Day Mega Sale at blinds.com.

1:13:53Nilay Patel:Rules and restrictions apply.