Ring's Jamie Siminoff thinks AI can reduce crime

22:33With LinkedIn Jobs AI Assistant, you can skip the confusing steps and recruiting jargon. It filters through applicants based on criteria you've set for your role and surfaces only the best matches so you're not stuck sorting through a mountain of resumes. Hire right the first time. Post your job for free at linkedin.com slash partner. Then promote it to use LinkedIn Jobs new AI Assistant, making it easier and faster to find top candidates. That's linkedin.com slash partner to post your job for free. Terms and conditions apply. Support for the show comes from Rippling. If you're a business owner, here's the truth.

23:17SaaS promised to make work easier. But now the average company is buried by hundreds of apps that silo your teams, slow you down, and simply don't work together. That's not SaaS. That's SAD. Software as a disservice. That's why you need Rippling. Rippling is the unified platform for global HR, payroll, IT, and finance. They've helped millions replace their mess of cobbled-together tools with one system designed to give leaders clarity, speed, and control. By uniting your employees, teams, and departments in one system, Rippling removes the bottlenecks, busywork, and silos your software created.

23:56Automated, perfectly in sync, and seriously simple to use, Rippling gives your company one source of truth for your people, their data, and everything they touch. With Rippling, you can run your entire HR, IT, and finance operations as one, or pick and choose the products that best fill the gaps in your software stack. And right now, you can get six months free when you go to rippling.com slash decoder. Learn more at r-i-p-p-l-i-n-g dot com slash decoder. That's rippling.com slash decoder for six months free. Terms and conditions apply.

24:39Welcome back. I'm talking with Ring's Jamie Siminoff about Amazon, selling his company at Amazon, leaving Amazon, coming back to Amazon, and managing his team and his product within Amazon's fairly legendary corporate environment. One of the more interesting themes that we experience making Decoder is everybody who comes into contact with Amazon leaves just speaking Amazon, right? Talking about one-way doors and two-way doors and two pizza teams and single-threaded owners. And it's very rare that Amazon's culture does not turn into everyone's culture. You had the opportunity to leave and come back.

25:19What parts of Amazon's culture were valuable out in the world? And what parts of Amazon's culture did you think, oh, this actually isn't working? When I go back, I want to actually reorient myself and tweak that. I think so. I do think the doc, like doing docs, it literally, I got out. That's the writing the memos and all that stuff. So I get out in the real world and I'm in like a real meeting with like people and they're doing a PowerPoint. And I literally lose my mind. Like I'm losing my, I can't sit here. I can't do this. I can't get information in this way anymore. Like I've been so trained.

25:54The thing about a doc that's so amazing is you get to like teach yourself the information. So you get to go at your speed. You get to think about things and sort of like you get to process it yourself. When it's on a PowerPoint, someone's literally like, they're teaching you at the speed of the entire room. It's all dumbed down. And so I realized like that, to me, the dock is one of these just like uber, uber powerful things. The one-way door and two-way door thing, I do think it's a great concept. I think it's been weaponized too much that it's too easy to say something's a one-way door. And so what I've decided is I came back that there are one-way doors, but you better not be able to break them down with a hammer.

26:35So like to me, it's like, don't tell me it's a one-way door unless it's really a one-way door. because a lot of the things that we sort of decide are, and for anyone listening who doesn't understand, like the two-way door decisions are like, it doesn't matter. You could like make the decision, you just go back on it. Like it's like a very sort of easy decision. One-way door decisions are supposed to be something that if you make that decision, basically you in essence can't change it. Like it's so impactful that you can't change it. And I think over time people have leaned too far into one-way doors as like, this is a one-way door decision and we have to meet on it versus like, no, it's actually not.

27:11Like it's a little bit painful, but so what? And so, yeah, my mental model is if you can break down the one-way door, it's a two-way door. Give me an example of your decision-making framework then. I mean, this is the other question I ask everybody on Decoder. Again, the joke is whenever I ask anybody who's come within 100 miles of Amazon headquarters, I say, how do you make decisions? And I hear about one-way doors and two-way doors. You've obviously riffed on that, right? But it is a good, it's a good mental model. So it does work because you should be allowing people at scale to make decisions that don't, that could be changed.

27:44Like that's, it makes sense, right? Like, like it's fast. Like, like if you want to make a decision that can be changed, it is not that impactful, make the decision. If it's a super impactful decision, we should talk about it. But it's just, but then it's like the, the, the bar on that. Like, how do you change the gauge to like how sensitive you are to like, what is a one-way door? That to me is, is, is what we have to do. And when I came back, you know, that's where we have to figure that out. And I do think I'm very good. Like my superpower is I'm good at making these decisions. I mean, I just, I was just in a meeting where you're branding a bunch of features and I'm like, do this, this, and this.

28:16And they're like, uh, it's like, do you, do we need to have like seven meetings? And I'm like, no, no, just do these. They're like, oh, you know, it's a kind of, everyone's like shocked. And I'm like, guys, it's, if it's a bad brand, we can just change it. Like it's a feature. Like it's like, I mean, but you don't want to do that. Like you don't want a call, it's like a new AI feature for us for one of our kind of motion alerts things. You wouldn't want to rebrand it. Like it's not great, but it's also like customers, like our neighbors aren't going to care if we go from calling it this to that.

28:45Like it's not going to like, it's not going to ruin their experience. What's going to hurt them is us not launching something or taking more time. Like that's worse. And so to me, I'd much rather make quick decisions than to sort of stew on these things. So I do think that one-way door, two-way door still is a good framework. I think it's just trying to figure out like what is a real one-way door. Do you think that that ability to move fast is the luxury of being the sort of the market leader ensconced in Amazon? I think it's the luxury of also being a founder. I do appreciate the fact that like the difference of being able to be the founder and having that, like, even though I am, listen, I'm an employee of Amazon.

29:20Like I don't, I'm not like trying to like act as if like I'm not, but there is something that you get from being the founder overall that I'd say you'd be harder to have if you just came in as a VP of whatever, you know, and you were recruited in. Yeah. That piece, right? The sort of like founder mode dynamic inside of a big company, Amazon has preserved it, I think, more than anyone, right? Like you mentioned Nick Weaver. I've known Nick forever. I knew Nick when he was selling Eero to Amazon. He retained the title of CEO. You had the title of CEO. You left, You came back. You told me right before I started recording.

29:55I actually did not. So I always had chief inventor. I actually never took CEO. Even when I was at Ring, I never had CEO. If you go back in my oldest emails, it says chief inventor and founder. Well, you get to do that when you're the founder of a standalone company. Of course. You can pick whatever title you want. But inside of Amazon's structure, you've purposefully not taken that title. Is there a reason for that inside of Amazon? Do you have all the authority of a CEO inside of that Amazon structure and you just don't have the title? I would say I probably do. I probably have what would be considered internally a CEO.

30:30I think it's a CEO's – it's a little bit of a misnomer, though, inside of a big company. Because the reality is there's a CEO. There's Andy. And then I definitely have the feeling, though, of autonomy to lead and make decisions for the area that I'm responsible for. So, like, whatever that's called, I do feel like I have that. I think Nick feels like he has that with Eero. And I think that's why people, it is not a thing. It is why people say, because you give them, whether it's a CEO title or, but it's, they certainly have the power to make the decisions. And, you know, in the end, what we like to do, what Nick likes to do, what, you know, founders like to do is build stuff and make stuff happen.

31:09And so that's the thing I missed most about being outside of Amazon and Ring is like, I can make things happen at scale here, which in a way that's just unbelievable. I mean, it's just like, we're launching this dog search party thing, which is like one of the most, I'm so excited about this thing because we're going to find, hopefully all these dogs. There's over a million dogs are entered into our neighbor's app every year. So it's like, the problem is so crazy. It's like, when we went, I said like, how many dogs are actually entered into our neighbor's app? I'm thinking they're going to come back with like, if they said 50 ,000, I'd be like, that's a lot.

31:42Wow, 50 ,000, that's crazy. They're like, there's over a million pet interactions a year on the neighbor's app. And my mind exploded. That's crazy. And so to be able to touch something like that and build something like that and think of something at that scale and get it out there and impact people, that's – I don't know. Especially for an inventor or founder, that's the coolest thing ever. I want to actually talk about Search Party and how that interacts with Amazon's platforms and Amazon's scale. I'm curious. There's founders. There's moving fast. I want to put some of the one-way door, two-way door dynamic into practice here.

32:16Okay. I look across sort of Amazon's portfolio, which is coming together. I look at Eero. Eero has a big bet on thread radios, right? Like my Eero routers, there's one sitting right over there, has thread radio in it. Yep. One day that's going to connect to Apple's thread network. And for some reason, the iPhone has a thread radio in it now. And this is like the smart home standard that a bunch of big companies, including to some extent Amazon, are pushing forward. For sure. You made a bet a long time ago when Ring was started on a different protocol, on Z-Wave, which I would describe as like the security system protocol.

32:47That's the one that all security systems run on. That feels like a one-way door. Like you made that decision. There's no coming back from it now to the big standard that your stable mate at Amazon has bet on. Blink operates on like a totally different random inexpensive RF protocol. There's a whole thing over there about why those cameras are cheap and can run on AA batteries forever. That's just like an Amazon, right? A big company has three divisions. They've all made three different technology bets. Someone, Andy Jassy, could say, what are we doing here? Make it all one platform. You didn't even make it as bad as it actually is.

33:20I also came out with Sidewalk, which is another protocol. So it's even worse than you're saying. So put that into practice. How does that work? Is there a meeting where someone's like, yo, we could get a ton more value if we undo what felt like a one-way door decision, and we all center ourselves on one platform, one protocol. For sure, especially those protocols, they get close to one-way door. The product that ships is the definition of a one-way door. You ship a Z-Wave product. That is a Z-Wave product. That's it. A Z-Wave product is a Z-Wave product. So that is a one-way door decision. The two-way door part, though, is if all of a sudden you see something in Thread that's really happening.

34:06The replacement cycle on these products call it is, you know, even, even on the long side, maybe three to five years, maybe six, seven, eight years. So if you look like if all of a sudden you saw that everything was really going to go thread and it was like, that's it. And if you were not on thread, you were like thread or dead. Like if that was like the, what we decided we could flip to that. We have enough, you know, of our toe in the water to figure that out. And so maybe you hurt yourself for like a tiny short-term thing, but you've also tried to figure out other things in the long term. So I think that one, as much as it's one-way door on the actual product, I think it's still two-way door to figure that out.

34:43And I do think like Sidewalk, I think we're going to see a lot of interesting stuff with Sidewalk next year, which is a sort of a IoT sort of protocol, but more really a replacement of internet for the trillion devices that are going to need to come online that have very low data. So very low, low data, like, you know, kind of in and around the home devices. The connection I make there is, you know, you look at the first ring cameras, look at the door bot, you had to invent a lot of stuff, right? There's chapters of your book that are, how am I going to get a camera to run on this power draw in this wifi enclosure?

35:21Whoops, I made the thing out of aluminum and I shouldn't have done that. There's a lot of that for the people who are interested in that, right? We are way on the other side of the smartphone commodity supply chain. You can just take a bunch of sensors and Wi-Fi chips and camera modules off the shelf and make a door. And you have lots of competitors who are effectively doing that. The innovation is going to come from, okay, AWS exists, right? There's an entire AI platform for us to build on. Alexa Plus exists. There's an entire AI platform. There's Echo devices. There's whatever. And I'm just wondering how you think about the balance of the things you want Ring to do versus the potential benefits of taking advantage of Amazon's ecosystem scale.

36:05Because that feels like the core tension of sort of the entire Amazon device ecosystem. For us, we're lucky that most of our products, and we have like a lot of different products, but if you look at the core products, they go to the cloud. And by going to the cloud, again, if you want to look at the two-way door, one-way door thing, the cloud's a two-way door. The cloud, you know, if a new NVIDIA chip comes out, it'll be available in the cloud. And so we don't, like, we're not that constrained because we're not, again, most of our products, especially our core products, they're not sort of stuck in the home where we can't upgrade them.

36:37And that's where we've been, you know, the older Ring doorbells are doing smart video descriptions. Like, they're not doing it because we planned so far ahead, you know, eight years ago, I wasn't smart enough to put an AI chip in it. It's because now there's AI chips in the cloud and they're already up there and that's what we use to do it. So I think it is, specifically for around us, like integrating with Alexa Plus is easy because it doesn't have to be done locally. Like that is the problem is when you get into these locally things, did you plan ahead enough? And that's the planning ahead is years ahead.

37:08Because if I think of a product now, the reality is on most products, it's only maybe one in six months, But it used to be it would take you like almost two years from when you thought of a product until it came out in the market. Then it's another year to get it to scale. So it's basically three years from when you think of a product until it's at a point where it has enough in the field to matter. And so whatever chip you chose three years earlier better have aged well because it's out there now. Moving everything to the cloud, particularly video footage of people's cameras from their homes, right, that's where the privacy concerns come into play.

37:43That's where the, hey, are we building an accidental surveillance network comes into play? You know, I'm just looking at the headlines in my prep talk. It's like you left. Amazon said we're going to stop working with police. You came back. Boy, Ring is going to work with police again, right? You have a partnership with Axon, which makes the taser that allows law enforcement to get access to Ring footage. Did that feel like a two-way door? Like you just made the wrong decision in your absence and you came back and said we're going to do this again? I don't know if it's wrong or right. I think different leadership does different things.

38:12I spent a lot of time going on ride-alongs. I spent a lot of times in areas that I'd say are not safe for those people. And I've seen a lot of things where I think we can impact it in a positive way. And so we don't work with police in the way of like, it's, you know, I just want to be careful. It's like, we're not, you know, what we do allow is that we allow agencies to ask for footage when something happens. And we allow our neighbors, which I'll say in this point, customers, just to be clear, we allow our customers to anonymously decide whether or not they want to partake in that. So if they decide they don't want to be part of this sort of network and don't want to help this public service agency that asked them, they just say no.

38:56If they decide that they do want to, which by the way, a lot of people do want to increase the security of their neighborhoods. A lot of people do want their kids to grow up in safer neighborhoods. A lot of people want to have the tools to do that and are in places that are dangerous. We give them the ability to say yes and make that more efficient for them to communicate with those public service agencies and also do it in a very auditable, you know, digital audit. Like that's the other side is that today, without these tools, if you wanted to have, you know, if a police officer wanted to go and get footage from something, they'd have to go and knock on the door and ask you.

39:33And that's not comfortable for anyone. It's also, there's no digital audit trail of it. And with this, they can do it efficiently, but there's also an audit trail. It's also, it's very clear and it's anonymous. If you say no, you never have to say like, that officer's at your door or you have to say like, no, which you could, I guess, say, it's very strange. It'd be like a weird situation. If you say no on this, they don't even know that they asked you. I'm curious, you know, you mentioned the audit trail. Well, I know you're really excited about the prospect of AI to analyze huge amounts of video and think about all the sensors in your home coming together with AI to bring people more insight.

40:11That's interesting and certainly that's how you get to build Search Party. There's the other side of it that we cover at The Verge all the time now, which is, boy, people are using Sora to generate footage that looks like Ring video doorbell footage. and ring being a verb, the angle of the doorbell, all that stuff creates a feeling of authenticity, even though the footage is totally synthetic, it's fake. Are you thinking about that? Hey, we need to put in content credentials to ring doorbells before law enforcement gets them so we can verify this is real and not AI generated? Yeah, I mean, we're certainly thinking about that.

40:46We've been thinking about that. And I do think it's where we have to, in the end, this source of truth is gonna have to come from like a secure server because I do believe these AIs will be able to generate, you can see it with Sora. It's like, there's some videos where I just, like, I have to watch it 50 times to understand if it's fake or real. And at some point, I think, like, if that's where it is today, in five years, it's going to be better. And so the only source of truth will be from these servers where it's like where it was captured. So you know, 100 % that you're getting like the video, like you're streaming it directly from there and there's no chain of custody issue.

41:24So I do think, yeah, I think about it a lot. And I think we have to, we're going to all have to go into a world where like where you got the information is going to matter because there's no way for a human to determine. Even watermarks, I think are going to get, it's going to be very hard to out-watermark an AI at some point. Yeah, I'm curious. I mean, there's the whoops, we relate to thread standards decision, right? By the way, I think you're late to thread. I don't know if it's coming through. I think you should ring cameras and thread. I'm listening. You want me to know if that's going to happen?

42:00That's one kind of decision with one set of stakes. Whoops, we were late to C2PA, which is the content authenticity initiative standard. Or we need to make a better standard and there's going to be a format war. Or actually we think standards are not the way to go because AI could maybe fake the metadata too. and we're going to build an entire evidentiary system that requires the cops to come to our secure server that we will authenticate. The stakes of that are radically different, right? In a world of AI deepfakes where everyone presumes video evidence is the gold standard and it's coming from your network.

42:33How are you thinking about structuring that decision? How are you thinking about privacy in that mix, right? That means you have to store a lot of video from a lot of people for a long time. And then how are you thinking about how quickly you might need to implement it in a world where AI is changing that so fast. Yeah. So if you go back, the privacy thing is that our customers control their video. That's it. So I think you always have it that customers... Actually, but let me just complicate that. Just presuming we have to have an authenticated server, right? There's a crime in my neighborhood and I'm opted in.

43:01And we're going to say the cops can only get the video from the Ring server because that's where we know it's true. I might not be as in control of my video anymore. No, because not how it's built and not while I'm here. because the way it works is that you will decide if you want to or not want to share that video, which is your property, with someone. Now, once you share it, then it is up to us to figure out, to your point, of how do we share it? How do we make sure that the digital fingerprint goes all the way through? Or how does that chain of custody work of this video to make sure there's no fake in the process of it?

43:37I think this is why it is important to build these systems. It's going to be important. know. I mean, this is like where also government is going to have to step in. We're going to have to deal with this across the board because we also have video coming off of cell phones. And we have, so we do need to figure out how to build, and there's going to be companies, you know, Axon would probably be one of the companies. I don't want to speak for them, but, you know, would they have evidence.com? So to build these evidentiary systems to take in, because there's, there's, Ring is one, one part of taking in sort of data around call it a crime scene.

44:08Cell phone video is maybe even more still today than that. So how do you take that in? How do you make sure that it actually was captured on the iPhone directly and not tampered with between the two things? We're going to have to figure it all out. I think we do have to work together on it. And the AI stuff is pushing us to sort of have to do it. I am proud with Ring is that we have built it so that you can take it directly, keep it on the server so you can understand where it was, where it's from, where it was created. And we have that digital fingerprint on it and the audit trail of it. Are you having meetings about this problem and that sort of cadence and with the urgency is we got to get through PDP faster?

44:47Yes. I mean, we are definitely trying to figure out how to make sure that our videos are always a source of truth. And right now, I mean, it's how we do our sharing. I mean, you go to a link and you take it from Ring. And so, I mean, I think, but I think you're going to have to do that more and more, um, as this, this world is changing. Like you just not going to be able to see just because someone sends you a video doesn't mean it's true. Yeah. I mean that to me, right. Just because someone sends you a video doesn't mean it's true is as deep a flip of our expectations of photography and video as has ever existed in like the history of photography.

45:22because it's up until now, you would have had to spend$100 ,000 to make that fake video or something. There's some crazy amount to, if you wanted to make what you're talking about, yours are like actually produced it. And now it's just, you type in, make me a video that looks like this and does this. It's crazy. Then it has really flipped this on the head. So I would say to anyone, like this is where you do your research, like use even AI to research what something's coming from. I mean, you have to like sort of use multiple, multiple things to figure out what is truth and what is not truth. It's like, well, we have to teach, like, we have to teach our kids that too.

45:59I mean, it's just, that's the world they're going to be in. And we need to just, like, you need to understand that. We have to take another quick break. We'll be back in just a minute.

46:1611.58 Uhr, Mittagspause. Your mouth is loud as the bureau-and. And then the chat is on. Canteed? Like always? You look at the like always-mickering portion and think about it. We should have to go to McDonald's. For the Big McDonald's hunger, try the new Big Gouda and the Big Tasty Red Steakhouse with 100 % Rindfleisch from Germany. So long the time is ready, not to our breakfast.

46:50Welcome back. I'm talking with Jamie Siminoff from Ring. Right before the break, we were talking both about how Ring uses AI and how so much AI slot being loose in the world changes the way people feel about Ring footage. Then I had to know, how does he see all these AI tools actually improving safety or our lives?

47:10I'm asking all this because, you know, again, you told my colleague Jen Toohey that with Ring cameras, we can get very close to zero crime. That's the combination of Ring and AI, and that you can get much closer to the mission than I ever thought. Zeroing out crime with Ring cameras, there's a lot of steps there. Explain what you mean by we can get close to zeroing out crime. And also, to asterisk that, if you go to the overall, what I said, in certain situations. So, like, around neighborhoods, I said, if you have all of our different products in a neighborhood, I do think with AI that we can finally see a path.

47:39Where before, I would say with the mission, you know, we want to reduce crime in neighborhoods. Great, we want to make neighborhoods safer. Okay, sure. But it was like a forever mission. Like I couldn't see with technology how to get to a point that you were, I could see where you were impactful, but like you could never like sort of get. But when you put AI into it, now all of a sudden you have this like human element that AI gives you. I think with our products in neighborhoods, and again, this is like, you have to be a little bit specific to it. I do see a path to get where we can actually start to get to where like, yeah, we're take down crime in a neighborhood to call it close to zero.

48:15And I even said there are some crimes that you can't stop, of course. So it's not – it's a little bit – it's always crazy to say something like zero-out crime. But it is – it's a good goal to have. And I think that's what we're trying to do is take the goal of like how far can we go in affecting this now with AI. Yeah. I read that quote and I was like, oh, that's Jamie. I know that guy. He's very in character. I mean, you got to have – think big. But mechanically walk people through what you mean. you put enough ring products in a neighborhood and then AI does what to them that helps you get closer to the mission of serial crime?

48:52The mental model or how I look at it is that AI allows us to have, if you had a security, if you had a neighborhood where you had, call it unlimited resources. So every house had security guards and those security guards were people that worked the same house for 10 years or 20 years. And I mean that from a knowledge perspective. So the knowledge they had of that house was extreme. Like they knew everything about you and that sort of residence and your family, how you lived, the people that come in and out. And then if that neighborhood had an HOA with, call it private security, and those private security were also sort of around and knew everything.

49:28Like what would happen? Like when a dog gets lost, you know, you'd be like, oh my gosh, my dog is lost. Well, they would call each other and one of them would find the dog very quickly. So like that's kind of like, so how do we sort of change that and bring that into the digital world? Can I just ask you a question about that neighborhood specifically? Do you ever stop and consider that that neighborhood might suck? Right? Like I've – just like the idea that every house on my street would have all-knowing private security guards and I would have an HOA and that HOA would have a private security force.

50:00There's – you can easily paint that to dystopia. Right? Like everyone is so afraid that we have private cops in every corner and I'm paying HOA fees, which is just a nightmare of its own. I would assume you live in a safe neighborhood. I hope so, yeah. No, I mean, today, I mean, so like I go to, I mean, if you want, I'll take you to a place where people live that they have to, when they get home from school, lock their doors and, you know, stay in their house and they can't go out. But I'm just saying the model, like the model is everybody is so afraid that they have private cops. I think the model is that doing crime in a neighborhood like that is not profitable.

50:35And I think that you want people to move into another job. Like I don't think that crime is a good thing. And so I think, I don't know, but like, listen, it certainly is an argument to have. I do believe that I think safer neighborhoods allow for kids to grow up in a better environment. And I think that that allows them to be able to focus on the things that matter. And so that's kind of what we're going for. I just wanted to challenge the premise that like, the model is there's cops everywhere, right? Like that level of private sphere. Yeah, it's not cops. I think it's more that you have the ability to understand what's happening.

51:10It's not like – but, I mean, yeah, I think, listen, it's a fair statement, I guess. I think, you know, I want to live in a safe place, so. The model is there's a lot of intelligence, right? I know that's what you're trying to say. Yeah, there's a ton of intelligence. There's a lot of intelligence in your neighborhood. Maybe it's private security. Maybe it's not. What does the AI do? Does it just make the cameras smarter? It lets you do more intelligent assessment of what the cameras are seeing? Yeah, I think the biggest thing is that the AI allows you to – Right now, we just say motion detection, motion detection, motion detection.

51:42It's funny. When I started Ring, the book was fun because I got to go back and actually go through this whole story of how this thing came. Motion detection was an amazing invention. You're in the airport, and there's a motion at your front door, and you look at it like, wow, this is crazy. Now, with AI, we shouldn't be telling you about motion detection. We should be telling you what's there. When should you look at it? When does it matter? and we shouldn't be bothering you all the time. That's what I mean about this idea of these sort of like a security guard at your house, security guard in the neighborhood is like, there should be this intelligence in your neighborhood that can tell you when you should be trying to be part of something, but not always telling you.

52:21So it's not just like car, car, dog, person, person. It's like, hey, look at this. Like you wanna pay attention to this right now. There's a piece of that intelligence. I mean, again, we have ring cameras around our house. Boy, am I often told that there's a package at our doorstep. We just had a baby. There's a lot of packages on our doorstep lately. There's a turn where you can connect that to Amazon's database, right? And the package that arrives at my doorstep, you know it's in it. There's a turn where you can connect it to, I don't know, a facial recognition database, and you can tell me who is at my door.

52:54Once you start connecting these databases, the privacy implications start to sort of fractally explode, right? Now we know a lot, and maybe we know too much. And that's where, you know, when we hear from privacy advocates when we do our coverage, it's the connecting of all the data sources that actually reduces the sense of privacy that people might feel in their home. And I hear you that, you know, if you look at my neighbor's feed, it's a bunch of lost pets and then constantly people asking if other people have heard an explosion. Everyone wants to know. Like, what was that sound? But when you connect a bunch of those databases, right, when you connect particularly to facial recognition, there's a turn in the privacy conversation where the stakes ratcheted up really high, right, where maybe it's gone forever.

53:35How are you thinking about that kind of decision making? OK, we have a lot of intelligence in the AI. It's trivial for the AI to go connect to another store of information, right? That's a thing you can do with AI, especially at a big company like Amazon where you have lots of other stores of information. There's a line. What's the line for you? I mean, there is a responsibility, obviously, just to like build, you know, safe products. So let's just start with that. Yeah, we did announce facial, like we call familiar faces, but that's not connected. That's just for your, like, that's, you know, how your, I mean, your iPhone today, if you search your iPhone, it's crazy.

54:09Like search for someone's name in your photos and they're like, their face, their pictures come up. And so I do think there's a, there's a balance between not allowing technology to exist that should exist that helps people and gives them more efficiency, gives them safer homes. And then also obviously not creating to where you're going to like this sort of dystopian place that, you know, is, is, is that. And so I think that, that is like, that's the responsibility, but like what we're doing with familiar faces is we're just giving you the ability to say like, when my wife comes home, don't, don't, don't, you know, because it is like silly.

54:45Like, why do I get an alert when my wife comes home? I don't want it. Like, I don't need it. I'm asking this for a lot of reasons. But, you know, I look at sort of broadly what's happening with surveillance footage out in the world. And I'm not saying Ring is participating in this. I'm just giving you the example. ICE has facial recognition systems, right? And they are arguing that a positive match in their facial recognition system is a definitive determination of someone's immigration status. That's way out there. I don't think you're doing that. But you can get to, OK, we have facial recognition.

55:15We have a bunch of evidence coming off of Ring cameras. Because to make it really safe, you want to go from passive surveillance to active surveillance, right? That's what the studies show. Now we can just – the camera will literally identify the criminal by face and tell the cops this person tried to steal a car from this driveway. And that's the thing that would get you to actually zero out crime. There's a lot of risk in those steps. But if I draw the thread from what you're saying, it's all the way to the criminals won't come here because the cameras will know who they are and tell the cops.

55:46Are you willing to go that far? it's not about the cat i think it's also that the that the cameras will alert people in a way that's that you know if you know part of what made ring and what made neighbors safer with like ring 1.0 and i think we are in like ring 2.0 is that there was no presence at the home like like what how did people break into homes it's like they go literally knock knock burglars they go like knock knock like no one was home and then like it was at three o 'clock in the afternoon they'd go to the homes next door find a place that was empty and they go in the home ring allowed you to now all of a sudden, like when someone comes up to the door, like you're like, oh, I got emotional alert.

56:20Hi, like what's going on? And so it gave presence to the home. So it didn't, you know, I don't think you have to go as far as that real time stuff to like get to where we're talking about. I think it's more of like the anomaly detection and allowing people to make it so that if, you know, if someone comes in that you're aware of what's happening around the neighborhood, because right now there's like no awareness of what's going on around. And so I don't, I don't think it's as, as dystopian as where you're going, it's really not what we're building. And I do think we can impact things to a really high level in neighborhoods, which again, to the Gen 2 thing, I mean, in neighborhoods is what we're talking about, that with AI and what we're doing with a bunch of rings together.

56:59And I think even dog search party is a good way to look at it, which is like, how do these cameras come together for good in a neighborhood? How do you think about hallucinations in this context? Your competitor, Google, just launched Gemini. They can do a bunch of recognition of various SNES cameras. The immediate reports were, this thing is saying a guy named Michael is in my house. There's like no one in my house named Michael. There's a deer in my living room. That's obviously not happening. You know, Google's models are cutting edge. They're as good as anyone's. They're still hallucinating, right?

57:28Well, I mean, all I'll say is like we've had smart video descriptions out for a while. And I would say ours were pretty good. So I won't talk about others, but I think ours are pretty good. The idea is that these things drive to a human decision, though. It's not that they're autonomously creating some sort of decision. It's that they're telling you what to do. So hallucination that there's a deer in your living room, it might be annoying because you're going to now check to see the deer in your living room and you realize that it's not a deer in your living room. But I do think it's driving – that's the idea is that these things drive to a human, that they're not sort of creating some sort of autonomous decision-making cycle.

58:06Do you think the models are good enough to do all the things you want to do? I've certainly, no one, I think, in our, I don't think anyone's ever seen something like how fast AI is moving from a technology side. I've never seen anything where every few weeks or every month, it's like literally something comes out that has surprised us that's that much better than the next thing. Well, let me flip that question. Is there something you want to build where the models aren't good enough yet? Probably not. Probably everything is there from the models being good enough. The question is from a processing, like the cost of processing might be off.

58:42So it might be that like the cost to do that thing is still so expensive that it's not rational to do. So again, the dog search party I go back to is like, you know, five years ago, could you have built something like that? Sure. It just would have cost you so much money, like in resources, time to develop it, everything like literally next to impossible to execute on. Whereas today I'd say it was reasonable. It still cost us money, but it's reasonable to do that kind of a product. Do you think when you talk about zero out crime in a neighborhood, the idea that everyone in a neighborhood has one of those illuminated ring signs in their front yard, is that enough of it?

59:23That's a part of it. Is that just enough of a deterrent? Like the bad guys will know, like my face is going to get captured on video and that will be analyzed by an AI and something will happen. Do you have to do more outbound deterrence? I think that's a part of it. I mean, awareness is a big part of it. I think there's ways with lights also, like lighting to do stuff that's a big part of it. I think having just, you know, if all of a sudden someone comes outside because something's an anomaly, that's a big part of it. Like it doesn't have to be, it doesn't have to be some sort of like crazy thing.

59:54Like I think a lot of these, and that's what I was saying, is it's a lot of these little things add up to make that work. So when you think about, okay, we can bring it down, we can bring crime down a neighborhood to close to zero in a neighborhood. What are the sort of ratcheting steps, right? Is it everyone just gets the ring camera and your platform does all the work? Is it someone gets caught and they tell all their friends in jail that they got caught? Don't go to that. Like, what are the steps? I think it's really about bringing neighbors together for this particular thing. So it's about how you individually, and we've always thought about how like each house is its own node controlled by the neighbor.

1:00:33So controlled by the person. And that's like, I'll keep going back to that, which is a hundred percent is that your video is your control. Everything you're doing is in your control, whether you sort of want to take part in anything is in your control. Like that has to be the first layer of all of it. But then when something happens, do you want to take part in it? So if you get an alert that like this dog looks like this dog that's in front of your house, can you contact your neighbor? You can decide not to take part in it and then no one will ever know and it's fine. It's just like basically deleted or you can take part in it.

1:01:02And so I think that's how we can do things that can make a neighborhood into this like node where individual neighbors are all sort of on their own, but when things happen, they can work together as they want to. And you think the AI will just sort of accelerate the process of where we're working together? I think AI, it is like a co-pilot. It's their assistant. It's helping them to figure this out. Because again, if you're just getting every motion alert and if you have, you know, if you have eight cameras and you're just getting emotional alerts all day, you can't, you know, no human being can parse all this data.

1:01:30And so that's where I, that's, that's, you know, kind of what I was talking to Jen about is that I do think I see now a way to use AI to help feed better data to us, which allows us to make better decisions and work together better. That's really interesting because that's a vision of AI that, you know, it has a, it has a beginning and end, right? There's a huge amount of data. AI can parse that data better. The LLM technology today can make better inferences out of that data. And I understand that processing chain. Let me ask you about the other side of it, which is you're part of a big company that has a smart home stack that has lots of microphones and speakers in lots of people's houses.

1:02:10I went to the Alexa Plus launch event and it was so funny. Pano specifically gave the demo of like, show me when the dog came home. And then it showed footage from a Ring camera. And I went, oh, that's not Alexa Plus. Like Ring already had that, right? You're just asking Ring to show the thing from its platform that it already had. I see all of that. I look at the big companies, Amazon, Google, Apple. No one has actually managed to add the LLM to the Assistant in a way that works great. We've all shipped it. Everyone's shipping it except for Apple, which apparently had to start over. But even the ones that are shipping Gemini with Google, Alexa Plus, right?

1:02:50They're first steps. And the idea that you can orchestrate between what's happening, like I just want to talk to you about my kid just wants to ask you questions about space, which is all she wants to do with any of these tools. And I need you to turn on the light. And that is a very deterministic process that you shouldn't get wrong. Like that orchestration is very complicated and no one has really quite nailed it yet. When you look at that and how rings should interact with that, does that feel like the AI tools can do the things we need them to do. Because when I say there's a product I want that I don't think the models can achieve, that's the first one that comes to mind.

1:03:24The LLM-powered assistant that does everything that we thought Alexa should do when it came out in 2014. I would say I'm very hopeful that Alexa Plus is there in a lot of ways today. Obviously, the team is working on lots of new features as well. I do think the vision that they have there for it, both what's out and in the market as well as what's coming is like, I think it's that. I mean, I do think it's that. And it's also where, as you said, like Ring, you know, it's great. You bought it maybe for security and you had it for this, but like it also tell you if you fed the dog or not and remind you because it actually like, there's a intelligence behind Alexa.

1:04:02And in this case, I'll say it's like almost like a house manager where it's taking this intelligence, this digital data and saying like, I think you didn't feed that. it's like eight o 'clock. Like you didn't see the dog. Like, do you want to feed the dog? And it's like, no, I did. I did it over here. Like you didn't see it or, oh, you're right. I forgot to feed the dog. Um, and so I, I do think, I think that's what we want is I listen, I think we all want assistance. And so like, what I'm talking about with ring is obviously like, I'm more focused on this idea of like security assistance for your home in your neighborhood.

1:04:34And what you're talking about then is like, how do you get, you know, assistance in your home, like inside the home, inside the four walls. Well, I think it's much more like I understand we are generating vastly more video footage than ever before. And we have vastly more sensors. And LLM seems like a capable, if not always appropriate tool to manage all that data, right? And to get something out of it that's human readable, like literally human readable. I understand that. And I see that argument. I see why there's so much interest there. It's the next turn, which is there's just an always-on intelligence in my house that can understand all of these different systems and connect them all together and actually sometimes do very predictably dumb things.

1:05:18I need the lights to turn on. And sometimes go out in the world and book a concert ticket. Like all the promises that we're making. I'm asking you because I think you have more insight into it. I'm not actually sure that LLM technology as it's presently constituted can make that leap all the way. I'd say I believe that we have all the technology pieces together today. So like the individual, like if it was a construction site, like all the things are on the site, like the wood, the concrete, like all the pieces are now there. I think to your point of like building the building, like there might be some pieces like where it's like we have to figure out how to exactly put the concrete together with the wood to make it like to make it exactly work.

1:05:57But I would say we now have crossed over to where I do believe that all of the actual technology pieces are there. I think some of it might be even on the processing side. Like it might just be too expensive. Like to do what you're trying to do might be so expensive. You might need an NVIDIA N100 in your house. And like that's just like – but that NVIDIA N100 in a year or two, you know, it could be like 50 bucks. And so if that's the case, then you probably will have that. But I look at the roadmap for Alexa and so what I can see, and I do think we're going to have that, in essence, always on intelligence that goes from turning on the lights to a turkey timer or whatever, but also to doing very complex things like understanding that it's garbage day, the garbage cans aren't out because seeing it on this.

1:06:44you know i am i've been at home like it kind of does like more than like it does multiple turns of intelligence like a human would do to say like this this is you know can you look at this you should do this as if it was like a full-time that's kind of what i say with the what i say like a security guard at your house and in the neighborhood what i meant about it is like this this idea of this like intelligence that really knows you and the only way to get that historically has been like a not only a full-time person but someone who's been there for a long time because they have to know everything you're doing.

1:07:13And I do think Alexa in the home will be that, if not already is on Alexa Plus in a lot of ways today. Well, Jamie, we're going to have to have you back when that all comes to fruition. I can't wait. And when you announce the inevitable thread-powered ring cameras. I believe that promise was made on the show. We're going to call them NP cameras. I'll give you the - I was promised a seven-month development cycle on these cameras. I'll do four months. I got to get faster. The book is called Ding Dong, How Ring Went from Shark Tank Reject to Everyone's Front Door. It is a very dishy read. If you are the sort of person who loves reading about how hardware is built and how companies are started, I can't recommend it enough.

1:07:50It is, I believe, available on Amazon. It is. I self-published it through Amazon. I mean, it's like, what else could you do? I mean, I had to. I mean, it's just I had to. But it was a great – that was fun too. Yeah. What's next for Ring? What should people be looking out for? We just launched all our new 4K cameras. They're awesome. Search Party for Dogs is going live very soon. Lots more fun stuff coming. And can't wait to show people what we can do. All right, man. We'll have to have you back soon. Thank you so much for being on Decoder. Thank you very much.

1:08:24I'd like to thank Jamie Sinoff for joining Decoder today. And thank you for listening. I hope you enjoyed it. To like to let us know what you thought about this episode or really anything else, drop us a line. You can email us at decoderatheverge.com. We really do read all the emails. Or you can hit me up directly on Threads or Blue Sky. now's a great time to ask us any questions that's on your mind too since our end of the year special is only a few weeks away decoder is on youtube now you can watch full episodes at decoder pod and you can also watch clips on tiktok and instagram they're at decoder pod too it's a lot of fun if you like decoder please share with your friends and subscribe wherever you get your podcasts decoder is a production of the verge and part of the vox media podcast network the show is produced by kate cox and nick stat and edited by ursa wright our editorial director is kevin mcshane the decoder music is by breakmaster cylinder we'll see you next time

1:09:07Thank you.