French founders and AI priests with Maxime Germain

We have priests using the product for their church. Now that's a power user. Are people praying to Beside? Is that what's happening? He has so many inbound for the church. The Beside use case is actually amazing for him. It seems like everyone in tech is building for themselves. But there's one startup building for everybody else. The real economy, as they call it, who spend all their time on their phones, not Zoom calls and Google Docs. This week on Access, we talked to Maxime Germain from Beside, a new service that replaces your phone and messages app with AI. Everything Google Voice should have been? We go through it all, from the virtues of being a French founder to the rise of the green bubble and building for the real economy, not just tech bros. But first, Alex and I react to the week's hottest topics, from ChatGPT's Code Red to Instagram's Choose Your Own Algorithm and Netflix versus YouTube? The fake economy. That's how they see it, at least. Without further ado, this is Access. Let's get into it. Ellis, did you ever have a Code Red when you were at Snap? I think so. The night of the Sony leak, when all of Evan's emails leaked and Mr. Private was on full public display, I think that was definitely qualified as Code Red. Mr. Private? Oh, yeah. Oh, yeah. He didn't say anything back then. That's true. Yeah, yeah, yeah. So, yeah, Code Reds are a thing in Silicon Valley. Everyone's been reading and hearing about this OpenAI Code Red. It's got. You know, it's gotten a lot of headlines in the last week, and deservedly so, because it's an interesting story. It's so funny. It doesn't help when you call it that internally, does it? No. I mean, you're begging for that to be leaked. That is actually a thing, though. That's like these companies do Code Reds. There was a Code Orange before the Code Red, right? It's an escalating code system. And really, it's like a focus exercise. It's like, you know, especially OpenAI, they have a ton of stuff going on. Focus on the core, which is ChatGPT, because, as I and others, I've been reporting, ChatGPT is starting to have some growth problems because of competition from Gemini, but also the moderation constraints that they've been putting on the product where it won't say things that it used to say. And that's been hurting time spent. I thought there was a really interesting story, though, in the Wall Street Journal recently that shed more light on this Code Red, and I wanted to get your reaction to it, Ellis. Because basically, what it's saying is that in this internal memo, that Sam Altman wrote to the OpenAI employees, that part of the work they're going to be doing on ChatGPT in the coming weeks is a, quote, better use of user signals, which is apparently just using the data of how people are using ChatGPT to reinforce how the product works. And amazingly, they weren't really doing this, I guess, before. And this is actually, like, very contentious internally, where you've got these high-minded kind of, we're on the path to AGI, research people, and then all the product people that they brought in from big tech. And they're going to be doing, basically, I mean, I read this as, like, ChatGPT is going to get more engagement-baity. And where it always asks to continue the thread, where it's always like, can I do another thing for you? Like, we're only going to see more of that. It's only going to go more that direction. That's what I read in this story, and I thought that was pretty interesting. Alex, it seems like you've had a rough week. Would you like to buy the new Haley Bieber smoothies? Would you like to buy the new Haley Bieber smoothie at Erewhon with this coupon? That's a gentic, yeah. That's a gentic use for this. Yeah, I mean, I don't necessarily read it as, like, getting more sycophantic, since that was such a polarizing issue and probably a bad product experience as well, since it's actively lying to you. And on the one hand, you could call all advertising lying to you. But I think the way that I read this is just kind of like all the announcements they've been making about plugging directly in to different shopping services to let you check out, I think maybe it's just going to be even more personalized and proactive. I mean, I was using it yesterday to research camera lenses as the average sad dad does, and I got my first, like, Insta-buy checkout link through Etsy where you could complete the whole process. And, I mean, they just have this infinite trove of information about you and your questions. This is what's made Google so powerful all these years. It's not just the stuff that you look at, but the things that you have intent to know more about and buy. And I would just, I would just, I would just expect them to dial that stuff up a little bit. But there are people, I think this is a coming culture clash because there are people in these AI labs who feel very differently about this technology than the technology that underpinned Google and Instagram being able to do an instant checkout and have you buy that pair of shoes that you saw an ad for. Like, a lot of people inside these labs think they're building super intelligence, something like God. And there is a real, I think, clash there. That this piece was getting into. And I thought it was important where you've got Sam who kind of represents the classic Silicon Valley VC archetype, you know, Y Combinator startup, ruthless competitor mentality. And then there are the people who, you know, built OpenAI over the last 10 years up until the last couple of years when it was a huge consumer company have not thought that way. It's been a research project to build AGI. And so, yeah, I mean, should ChatGPT, be more personalized to be a better product? Yes. Or is it going to be tempting, I think, for them to use the persuasiveness of this AI to build one of the most engaging, hard to get away from products of all time? Yeah, I think that's probably the thing they're trying to balance. And what this code read is probably going to show on the other side is where they land. And I have some faith. I mean, with that cultural mix internally, while some folks are going to have a bad time, I think what's important is that that conversation is happening. And that with a leader like Sam at the helm, who seems to have some ideas of goals for society beyond engagement, I think I'm hopeful. I mean, he said on Jimmy Fallon the other day, Jimmy asked him, do you use ChatGPT when raising your baby? Sam says, I cannot imagine figuring out how to raise a newborn without ChatGPT. And I just like, you know, have to agree like a thousand percent, like, first baby up Googling content farms at 3 a.m. Absolute nightmare inside of a nightmare because everybody's telling you something different. ChatGPT tells you kind of the synthesis of everything that it's hearing. And if it wants to have a buy button under the internet's top five favorite strollers, I'm fine with that. So you're fine with like engagement bait, the hell out of it. It's fine. I mean, we're going to have to figure out the right, the right modality for it. I think what's good about it is that there's competition here, right? I think with Google, there's just so little competition over the years. They just made the shopping links or Amazon for that matter. Before you know it, three quarters of the page is shopping links. And so I also do think that like healthy competition, including from Gemini, which is getting top placement in Google and in Chrome to come. I think they're probably going to be maybe a little bit less shoot first, ask questions later if people aren't responding to it. Yeah. I did. I asked, we talked about it last week. Neural lips, that AI conference. I asked people there for a little lips where, which is basically the largest gathering of AI researchers in the world. It was in San Diego last week. I asked a bunch of people who were there to talk about kind of what people were discussing for sources this week. And Google was the ascendant. I mean, Google was the winner of for sure. Everyone felt like Google was, was at the top and opening and I felt more shaky. So yeah, open eyes definitely on it. It's back foot a little bit. I mean, they still have, you know, approaching 900 million weekly users, so they're going to be okay, but it is fiercely competitive. The other thing just to note is that yeah, GPT 5.2 coming out probably this week is what I'm hearing. I don't expect it to be a huge thing. Like that's kind of tit for tat with Gemini, but I think open eyes betting on this model release in January with better images, multimodality personality, tweets, speaking to be the thing that really gets them out of this code red. Um, and it's interesting how much they're focusing on personality. You know, it's like when you talk to Claude, you can really see the difference of how do that thing you do. Alex, say it, Claude. Um, when you talk to cloud, which I've been using, I've been using a Opus 4.5, the, the, which is an incredible model, but like they don't do this personality stuff like chat GPT. You can't like have it talk to you like a nerd or, or concise or whatever. If anything, like it's its own personality, it tells you like you're staying up too late and like, just stop using me. So there's like a real divergence in like the product strategies where open AI is going more of the Facebook route, I think, and Anthropix doing something else. And I think that's, that's good for the market. It's good to have that. The reason is an interesting question too, just in terms of like behavioral science and consumer psychology. Like while Google does have their AI overviews at the top of every Google search now, they can only apply it to stuff that you're Googling, right? And so chat GPT already just has this unbelievable Kleenex like brand recognition where people are asking deep questions about their life. I'm not sure how you train people to start putting that into the Google search box. So unless they're going to get you over to Gemini, I think they do have that top placement, but even for chat GPT versus the others, I do see this interesting tension. People ask me all the time about like, you know, is AI going to be horizontal or vertical? And the more I think about it, I think people are not going to be super super comfortable using one AI with all these different topics and all these different tones of voices about your life. Like I've asked it for advice sometime. And then I'm like, can we switch gears to like working on my manifesto again? And it's just kind of weird. You know what I'm saying? Like it isn't how you would treat a person. And that's, I think why a lot of people like have these hard walls with their therapist about the stuff that they do and don't talk about. It's because those tonal shifts and contexts really do matter. So yeah, I wonder what metrics ChatGPT is seeing. Are people switching away? Are they using it less? I wonder even where that code red came from. Because I mean, I have a lot of faith that they just have so much staying power. I know exactly where it came from. Because I've been reporting on this. They dialed back how sycophantic and rabbit holy the model was in August. It started tanking time spent. I reported on a call where they acknowledged this with investors a couple months later. They did code. Orange and like October, Gemini three came out in November, they do the code red. And now they're in code red. So it's a combination of like self inflicted pain, where they're trying to balance like growing with the bad externalities of that growth, and more competition and Google like Google waking up. So yeah, it's a it's a it's a hard time for Sama. This episode is brought to you by Accenture. When your advertising operations fall out of sync, everything else follows. Spotify and Accenture are working together to reinvent the rhythm of ad sales using automation, analytics and smarter workflows to simplify campaign delivery and access better data across the business. The result, less time spent on operations, more time connecting brands with the moments and fandoms that matter most. Learn more at Accenture.com slash Spotify. This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50 page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required compatibility and availability varies 18 plus. I think this is actually the perfect segue into this next topic. Like how much control do you give users over how much junk food they want to eat? Tell us about this, this AI, this, I was going to say AI story. It's an AI story. We can call it an algorithm story. The old, the old days of the algorithm where we only talked about algos. Well, I just thought it was interesting that this week, Instagram and Spotify had these announcements about basically making it your algorithm, so in Spotify's case, it's, you know, you can prompt play playlists, which they've been doing versions of, but I think the end state of what they're talking about is like you being able to dictate to Spotify what it recommends you. Uh, and then Instagram basically doing the same thing with the reels algorithm where you can prompt it and change it. And it'll also show you what it thinks it knows about you. And you can share that they're making that like a shareable social thing. Um, where, so these, these, these, these models that power these feeds that really like, uh, intermediate our lives in a pretty profound way that we don't think about. They've been very like opaque for so long. And I think with all these platforms plugging in LLMs to redo their algorithms, they're like, Oh, what if we just like chat GPTify this? And in the same way that people prompt the hell out of their chat bots and their assistants, like why would you not just prompt your algorithm to be the thing that you want it to be? Right. Yeah. I think natural language settings for stuff like this is the bomb. Like being able to ask for things within an app. Like one of the examples Spotify used is that you could say music from my top artists from the last five years and then push it further with feature deep cuts. I haven't heard yet. Like that is ostensibly something you could maybe hack together using like smart playlists and Apple music if they even still work at all. Um, but that's cool. Right. Being able to use natural language to query that database. Um, the th the thing though, that just like really messes me up about this Spotify announcement. They said, imagine a Spotify that doesn't just passively learn from you, but literally listens to you. And it's like, man, is this not like what we've been asking for from tech for the last 15 years to not just look at our clicks, but like, listen to us about how we want to spend our lives and, and, uh, what feels good? Because we all know that when you're on the toilet, like, yeah, it's awfully easy to just click on whatever's most engaging. Uh, even if you know it's bad for you when you're in a bad mood and you walk into the convenience store, it's so easy to just grab, you know, whatever sweet treat you want to grab out of the aisle. And they've just made it so easy with so little control to have influence over it. And so I'll tell you my conspiracy theory, Alex, I would bet you that this whole Instagram thing where you get to choose your topics, I would bet you that at most this impacts like 30% of the algorithm. Like, I don't think they're willing to give over full control and let people say, actually, I don't want to see any more hot bodies. Actually, I don't want to see any more camera stuff. Cause I always Insta buy it. Actually, I don't want to see, you know, like, I don't think they're willing to turn that over though. I do think this is a great step and I'm, I'm honestly shocked that the, that they're doing it. I wonder like what it says or what it means that, that they're, that they're doing this. If anything, there's probably like, Regulation in the EU or somewhere that's like forcing them to do this, you know? But who knows? I mean, it's, it's cool. I'm excited to play with it. I did want to talk briefly about the big media news that everyone's talking about. I mean, it's everywhere. The Netflix Warner bros or paramount thing. I mean, it's straight out of succession episode. It's straight. Yeah. You've got the, see, we're completing each other's sentences and we're not even at episode 15 yet. Yeah. The, uh, the Ellison's, uh, reminds me a lot of the succession family. Um, but I just think it's been interesting as like, we live in LA where, but we're, I guess, quote unquote, tech people to hear the way Netflix is trying to defend this deal. And they have to find a way to defend it because it is, it is very bad monopolization of Hollywood, right? You're going to have the number one streamer combined with the three or four streamer. Uh, and just, you know, there's going to be layoffs. There's going to be cost savings as they call it, right? It's not going to be good in the short term for Hollywood. So I totally understand why people in Hollywood are against this at the same time. I also don't understand why Netflix is going around to regulators in the press and going like, we have to do this because of YouTube as if anything that people do on YouTube actually competes with the kind of time spent that you're talking about when you're watching Harry Potter, right? Like is it maybe, maybe the world is that zero sum and maybe growing YouTube usage is actually eating into that. And Netflix thinks that adding the Warner Bros and HBO library will somehow steal time spent from YouTube. But I just don't see it. It's like, it's like comparing or like Air One and Costco or something like these are just totally different things. Like, yeah, you can go get groceries at both, but it's a totally different kind of customer. It's a totally different reason you go in the store. Well, can you summarize that case? Can you put that in a sentence? We need blank to compete with YouTube. They're basically Netflix is making the argument that they need this content, this extra content and scale Potter or, well, yeah, I mean all the Warner Bros and HBO stuff because their real competition is not the rest of Hollywood. It's YouTube, which is what you say when you've conquered Hollywood. Like there's no one left in Hollywood that competes with Netflix at that scale. It's just like Disney after them. And they have not, you know, a hundred million or so fewer subscribers. And so if you're in a Netflix position, it kind of reminded me of when I was in the courtroom for the Meta antitrust trial earlier this year. And they were talking about like their competition is time spent across the internet, right? It's not quote unquote social media. Every big company, when they're trying to defend something that they're doing, that's inherently bad for competition. We'll just point to the next biggest thing. But yes, maybe there's an argument that YouTube and Netflix are competitors. I can see that. And maybe that does correlate to the time I spend on Netflix. But, um, I just think people who work inside Netflix and YouTube know that like the reason people come to their platforms are very different. I just don't see how that's a lever to even improve on that advantage. If that is your competitor, you're like, we have the biggest content library and we need more in order to compete with YouTube. I think if anything, there is a medium difference. Like if they're like, we need to acquire the world's biggest YouTube content owner or agency or whatever, like, it's a completely different habit, a completely different, uh, device most of the time, though. I know YouTube is blowing up on TV. So maybe, maybe my, maybe my air one Costco analogy is not even right. It's like the real analogy is like, like Netflix is, is Costco and YouTube is like going farming and farming your own groceries. Like these are just totally different things. Like YouTube is UGC. Um, and Netflix pays a lot of money for its content. And to say that like, yeah, there are, and you know, finite amount of hours in the day and people choose to spend that time but youtube is not you know doing um you know k-pop demon hunters like youtube's not doing the kind of stuff that's on netflix so to make the argument that like you need warner pros to compete with youtube i just i think that's actually funny because at the hamburger house and we'll close here at the hamburger house if you can get a picture of that in your mind we watch the like netflix k-pop demon hunters and then we switch over to youtube to watch like every goddamn permutation of those songs and so we are successfully using both platforms let's leave it there that was some good news uh after the break we will have maxime from beside we will see you there this episode is brought to you by accenture when your advertising operations fall out of sync everything else follows spotify and accenture are working together to reinvent the rhythm of ad sales using automation analytics and smarter workflows to simplify campaign delivery and access better data across the business the result less time spent on operations more time connecting brands with the moments and fandoms that matter most learn more at accenture.com spotify this episode is brought to you by google chrome you think you know a browser but gemini and chrome that's new it can help you with practically anything on the web like restoring a vintage motorcycle from a 50 page restoration block or finally break down that long article you've had open for weeks gemini and chrome is here for it ready to make anything online make sense there's no place like chrome check responses set up required compatibility and availability varies 18 plus actually i like the idea of max just starting it max can you just say what you said again so i you know i prepared a bit uh and we sort of have a backstory uh with ellis uh because we've met uh twice in real life maybe three times but two times that i remember one time uh you know i was very young i think i was 21 or 22 uh and we met uh at this crazy party in new york and ellis was still a journalist at the verge at the time and our friend took us uh club in new york uh called the box where there was me ellis uh brian chesky was here with us as well it was a small group of like early consumer tech it was really fun um it's called the box and the entire looks like that and this is exactly the view me and ellis had uh when we were there because we were just sitting on that balcony up there uh and the second time ellis i don't know the name of the place i don't know the restaurant but we decided it was going to be the classic move when we come to uh la and it's uh you know the picture is from that place that we went to it is what was brian chesky doing with you guys was he were you all at the same table like what was the no so i think we had the club for ourselves because fred decided that he was going to be a private party uh and i think at the time fred was hanging out a lot of times with us with brian brian happened to be in town and he was in new york so i had to go and get him at the door with fred because he didn't know if he was really the right place that he should go into so we came and picked up brian at the door to get him upstairs at the private party with all of us which was a crazy crazy night i think that was like the latest i ever stayed up as a reporter i was like oh this is clearly a special night i'm gonna be up on a school night until 4 a.m but it's going to be completely worth it i mean i'd never been someplace that i mean like that i mean this is just also how much money there was in like the app store boom of like 2010 2011 2012 i mean there hasn't really been anything like that since and while i do think the culture has changed uh that was definitely a special night and i guess i guess you weren't yet uh on our level max what what were you doing at the time i was you know part of the the french mafia crew of early uh you know french founders building consumer app in new york i was hanging out a lot with pierre valade from sunrise fred who had phone back at the time i was running a company i dropped out of school when i was 20 i was running my first startup uh from my office in tribeca uh and on the way back home with my red backpack fred calls me and tells me you gotta come you gotta join this party so i went so that's how i ended up there randomly shout out here you can answer how come all my favorite product founders are french this is something i've been wanting to know for a very long time and you just mentioned a few of them well i do think uh i see you with a very european soul in many ways uh so i think there is like there's probably in the european culture uh more of an emotional connection to the to the art to great product um there is a more more depth in understanding uh you know psychology of people and getting to understand these things and i feel like a bunch of french founders even those coming with a very technical background still have uh uh you know this sensitivity uh and the ability to sort of like translate into tech to emotions to consumer which worked really well um and yeah i do feel like that's something that i've always seen you appreciate in the way you appreciate product in the way you talk with people even in the the work you do or what you're really good at uh i do uh you could be you could be european you could be french uh in many ways it's the best compliment i've had all month i think this actually reminds me of that night at musso and frank's and i did want to ask about this you know the difference between uh american founders and european founders or french founders especially in ai but i remember that night at musso and frank's which is like a big steakhouse in la that's kind of like old hollywood remember just how like pumped you were to be there you're like everybody's hustling everybody's talking about business and you're like no one in france wants to talk about work over drinks and i just like never thought about that i mean is that still true um yeah it's super true i think like once like people people have work-life balance uh and they sort of like compartmentalize in life and when work is over they sort of like get into an other mood where like you don't really want to bring work into that that place which i think can be great uh sometime to disconnect but also does not foster an environment where you know you're pumped by what you do you're inspired by the people around you there's kind of like a big downside to it because i don't think it's necessarily about business the thing we want to talk about i think we want to talk about these ideas we want to talk about building and we want to talk about business and we want to talk about the craft all of this is not only uh about making money or doing these things and i feel like once you have uh you know i guess like all the world that i that i evolved in there's a lot more uh passion around what we do so i kind of have that desire to talk about these things not because i think it's work or it's business but more because it's completely part of life and there's so many interesting things going on doesn't necessarily have to be compartmentalized so every time i'm in the u.s i'm like reminded that uh there we i'm gonna have tons of great discussions with people who are actually going to be optimistic and excited about the things that i do so i just want to hear ellis talking a french accent more that's all i want to hear at dinner i think it started off well and then it took a turn for the worst no you were uh you were uh on track i just love it compartmentalize you think i'd speak uh with a better accent knowing that i lived for 10 years in new york but apparently i can't no why would you want to talk like us man it's so boring well do you know about california english maxine it has more like vocal fry well yeah but also there's just like this whole meme online like if you say like yes in california english is yeah no pretty much for sure can you try that in that accent all right but in in return you're gonna have to pronounce my name properly okay okay you go first yeah no pretty much for sure for sure bro okay so now my turn uh maxime germain germain germain well you typed huh h-u-h yeah because i was trying to put you in the right track this is good yeah but i kind of like maxime germain because as i told you it looks like feels good like if it sounds american to french people it sounds cool on that on that thread i mean how do you feel as you're speaking french i mean i'm a european founder right now i feel like there was a lot of momentum in the early ai boom whether it was with mistral or other ai startups in europe and i feel like i'm not hearing about quite as many i mean what is it what does that feel like right now what are the differences i don't even look at these things to be honest or i don't even consider us being european founders or no like we run a u.s business have tons of people in the u.s i spend a lot of time there as well um i guess there was an early there was a lot of noise on but i'm not sure you know noise translate to actual value created or that hype is actually the right driver you should be looking for so I always try to, and you know, we're in stealth for 18 months as a company, like even two years. So I was trying to avoid that. And I don't participate in the scene at all in Paris. I might have like a couple of founder's friends, but I'm mostly just working. So I'm kind of like out of that. I don't really have an opinion, if I'm being honest. It is especially funny because do the grandmas and real estate agents that you're now serving with Beside know that there's some high fashion French guys building their favorite product? I hope so. I hope they know. I hope they feel it in the product. Because I don't know if that's ever been done before. Well, you know, we do speak with them and it's kind of, it's very interesting because I come from that world in many ways. I was born in a small town in Le Mans. All of my family work in the real economy. I don't come from the tech world. I don't come from, you know. I come from a very, you know, knowledge worker background. So, you know, it doesn't feel super unfamiliar to me because these are the people I grew up with that my first business was actually building websites for plumbers and for, you know, electrician and all the local businesses around me. So I just feel like that has just been a full loop to bring that back to the business. I love when we call it knowledge work. It's like you don't need knowledge to be a plumber or to go out in the real world and do something. It's super demeaning, honestly. Like it's just become part of the vernacular, but it is kind of crazy when you think about it. That's why I was looking for my words. I couldn't find the right way to put it. But yeah, I agree. Yeah. And in full disclosure, Max and I have worked together for a couple of years now. He was actually one of my very first clients at Meaning. And now I'm advising. And one of my favorite things about it is that Max and the crew are bringing this. Their level of thoughtfulness and even just like beautiful, beautiful UI design and brand design to an audience that doesn't usually get that type of love. I mean, I feel like when you look in the news and AI every day, it is almost exclusively stuff for other people in tech. Knowledge workers. Yeah. How did you arrive at this whole real economy idea? And do you feel like they need it more than we do? They need the AI more than us because it certainly feels that way to me. Yeah. Well, so, you know, multiple things came together at once. One, you know, there is this thing from Virgil Abloh that I love who was saying, like, one of the best way to do design is to actually only tweak 3% of an existing product. And the best ideas are oftentimes a 3% tweak of an existing design. So the way we were considering this. Internally was that the product should be and feel very neutral to the daily drivers that they already have. So we really took iMessage as a base for the product. And we felt like if we were going to bring AI into these products, we should tweak the 3% of the UI of iMessage to bring that AI layer in it. Because if you're working off something that's already. Yeah. Widely used and adopted, you're already working a lot of very familiar user experience patterns, user interface patterns that are known. And I think every time people are trying to build their own tools or have, like, tons of opinions in the way things should be on 90%, I think it's too much. But if you go down to tweaking the 3%, you actually. Are able to drive a lot more impact for the end user. So that was kind of the philosophy where we wanted to keep the product super horizontal and the design extremely neutral, taking a product that has billions of daily active user like iMessage and tweaking the 3% there. And I think then the thesis we had for the company was not that we would end up with product market fit for the real economy at all. I thought I was building for myself. Initially. But I kind of like really like this non opinionated way of looking at if we bring AI into conversations into a mobile product where all of your conversations would be recorded, where you would have, you know, sort of second brain from all of this conversation where the AI could answer phones for you, the phone for you or take action based on that conversation. So I think that's a really good point. And I think that's a really good point. For whom would it be the most valuable? And we reversed that before launching the product. We started by launching ads, talking about the feature set without saying for whom it was. And what we learned as we were building the V0 of the product was that it would likely be successful with people working in the real economy because they were the people the most excited about the feature set we were building. And even the way we keep building, we build in the less opinionated way as possible. And we completely surrender to the signal we get from the market and from the users and from usage and all these different things. What is the real economy as you define it? Well, at the moment, it's people who are not working behind a laptop, but for whom conversations is where business happen. So you can take. You can take that in many ways because the product market fit that we have today, the number one user using the product 250 times a day is an insurance group. And it's the person answering the phone for an insurance group who's always on the go. They're always on their phone. They need the context of their conversation. Now, that's a power user. Power user. Hold on. Hold on. Are people praying to be so? Beside? No. He has so many inbound for the church and so many things to remember about the schedule of people, the things they talk about and like all this stuff that he wants this conversation to happen in beside. So he can offer back an amazing, meaningful experience when he has a conversation with all of these people reaching out. They're often reaching out for help or for, you know, personal matters. So the beside use case is actually amazing for him. Do people know that? It's not him because you can train with your app. You can train it to be in your voice, right? So you have two things, right? You have the AI receptionist. That is one part of what we offer. The second part is all of your phone conversation are actually recorded, transcribed and summarized. So what we've done is that we've rebuilt, you know, phone and texting from scratch. One break is the AI receptionist, but that's only one of the break. You then have, you know, full on recorder and transcriber on. So sometimes it's actually just him picking up the phone and not the AI receptionist, but the AI receptionist that he filled, he filled with tons of data from the Bible and like specific things that he wants, you know, the receptionist to say, which is one of the crazy use case we have. But there are tons, you know, like weird use case in the product that we see. Well, it's not in his voice. So, right. You can do voice cloning, but for him, the use case is him talking actually. Yeah. It's, uh, it's really interesting to me because I think sitting at our desks every day as tech fans and builders, you kind of just assume everybody's a little bit like you sitting at a desk. And when, you know, all these different meeting recorder companies come out, you feel like it's commoditized. And then you realize it's like, wait, there's all these people who aren't using Slack or Google meet or zoom. They're literally. They're literally using their phone. And I think that's one of the other cool things that you've done is that so many companies in AI these days are building wrappers, which may be fine. If you build the right UX or UI for a specific vertical, you fine tune it, et cetera. But you also have that actual technical innovation of like, as you said, rebuilding the telephony stack, you know, allowing people to get a new phone number. Can you tell us about, about building that in, in the year 2025? Because as far as I know, nobody's tried that in a minute. Uh, but maybe, maybe I'm wrong. I feel like that was always kind of the Google voice dream or something like that. And then it's just kind of fell by the wayside. Yeah. Uh, well, uh, I think there, there, there are two things to that. One is, um, where you're right is you kind of want to be able to, if everything's going to be commoditized and if the models are only going to get better. And if we think about the context of AGI and on the way to AGI, what do we need, uh, to solve, to actually be able to deliver what we all think. Uh, AGI can be, I think there is a big part of that. That's going to be the life context of your life, uh, everything happening in your life. And one of the contexts that is, uh, hard to capture is conversations. Um, so on MacOS or on your desktop, it's a lot easier because you can record at the OS level and we've seen all of these bots coming, et cetera. But when you come to the mobile phone and when you come to trying to record a phone conversation, it gets quite hard because you don't have access to it. by default on iOS or on Android. So you really have to be able to power entirely the phone number and all the infra going behind that phone number. Now that we power, I think it's like 2.5 million phone calls every month, we're talking to the infrastructure provider, we're working with them, we're talking with the right lawyers in the space to build all of that. And so, yeah, to power that at scale, it's a team of five people in the company. It's probably the biggest moat that we create because both on the regulatory side and the technical side, it's extremely difficult to build. How excited are you to sell the company to Verizon? No, they're looking into that actually. Many of these, they are all reaching out to run pilots with AI companies because they understand some of these things. And they actually want to see what kind of AI feature we can offer to some premium plans for these people. I find it to be super interesting. Yeah, I feel like in a lot of ways, what you've built, especially being inspired by iMessage and the phone app, just reveals how much there is that Apple and Google just haven't really been doing or thinking about when it comes to actually using AI to help us. Do you expect them to come into this market sooner rather than later? Do you feel like they'll kind of let the application, the applications take care of this? Well, Apple does the iOS. I mean, in the new iOS, you can have Siri answer the phone for you. I did it the other day, and I think the person just hung up immediately. It's pretty simplistic. It's pretty simplistic, right? So they're starting to edge into it already. And you can do a call recording and get a transcript on it, but it shows up in the Notes app, which is blasphemously kludgy these days. So yeah, obviously they're not experts in AI either. Yeah, well, I think, it's wasted potential for both of these companies because there are absolutely amazing experiences to be created if you control the OS with AI. I feel like it's kind of like a widespread feeling in the industry right now where we're all wondering what they're up to and why hasn't this gotten any better. I do think that they will catch up. I do think that I don't see a reason why Apple or Google won't get even better. Actually, Android is a lot better than Apple on the AI, at the OS level. What we're discovering though is, I think like, and one of the product theory that I always have is that I think it's easier to grow into usage than to start from usage. So if you start building with an OS that already has a billion daily active users or 2 billion, and you're trying to build these features, you're trying to build them to please a billion or 2 billion, you're not growing into that base. And as soon as you push something, it's like, for everyone. So how much depth can you really cover when you're that horizontal? Can you actually build for specific workflows of people? One of the insights we have in the product is that that context is leveraged to be shared with teams. I don't think Apple went into making even macOS or iOS functional as like a business tool or that they would bring the concept of teams within what they build while it is the core of what we offer and becomes the core value as we progress into this unopinionated way of going about things. That's what we keep uncovering is that the shared context within the team, the collaboration features are the key to grow the product further and to expand to the product market fit. So even though they could and they probably will add more AI in the details, of their execution and in the details of where we're growing with the product market fit, I don't think it will actually merge or become too competitive that we don't have a shot at existing. I think it's going to be a big deal. Like restoring a vintage motorcycle from a 50 page restoration block or finally break down that long article you've had open for weeks. - It's a good point about not just AI, but about built for work is something that Apple at least is not, not quite as good at. I mean, it's just funny to imagine that everybody at Apple is like, are they using Slack or are they just using iMessage to communicate? Are they using Pages and Keynote? I believe that they are, which have pretty, you know, ancient collaborative tools and I don't even think they have comments. And meanwhile, you know, definitely a little bit of a different situation for Google where they make so much of their money from a lot of this stuff. But to your point, I mean, even with Gemini being added to Chrome, it's one more button amongst 50 buttons. And even what you were saying about the Virgil Abloh quote, I mean, it just reminds me of my time at the browser company where you want to do so much because you see so many problems and so many opportunities. And so you create this incredible browser with all these new things. But at the end of the day, if people can't get used to the vertical sidebar, it's kind of a non-starter because you went too far over that 3%. And yeah, I mean, I'm curious, like for the people who are using Beside, like, what is the magic moment that is most valuable that hits them quickest when they start using it? - The onboarding for a business is consumer grade, leveraging all the best practice of all the top consumer product building where we know how scarce attention is during onboarding, how quickly people can drop and how easy we have to make that. So we're able to get people to experience the core value of the product, quite fast on their own data by doing one, a really good website scrapper. So we're able to extract all the information about their business, pre-fill that AI receptionist during onboarding. So when you finish the onboarding, you're actually having your first conversation with the AI receptionist that already has data on your business and can already answer quite a bunch of questions as if you were a customer. And the second thing- - So you just add your website and Beside scrapes, all the information for you that's publicly available and uses that to build the bot. - Exactly, yeah. - Exactly. - Yeah. And then we call forward your phone to Beside, meaning that your existing phone number that you've already adopted actually redirects to Beside. So that's sort of like, you're gonna have all of your next calls within the app. You're gonna discover the transcriber, you're gonna discover, the call summary, you're gonna see all these things from there. So these are the two tricks that leads to adoption in the easiest possible way in discovering the core value prop of the product. - I was gonna say, does this force you to be a green bubble though, Max? I don't know if you could put a price on that in terms of the barriers to entry and retention. - Yeah, we're a green bubble, at least. It's a green bubble world. There is no- - There is no blue bubble royalty potential in this. And we talked about it earlier. I remember you made me freak out with that blue bubble world early in the company. Turns out it's actually fine. People are, okay, I haven't received a support message on the blue bubble. - Well, how much of the real economy, I mean, I bet you the ratio is quite different for Android versus iPhone. - So the people getting into the product app, often, the business owners, they're all on iPhones, but we had to build Android because the moment they want to invite the team, you enter that like 50/50 split where it's Android, iOS 50/50, yeah. - So if I use Beside, if I really use it, I am not sending blue bubbles anymore. - You're sending green bubbles. - Okay, interesting. - I think the idea though is that it would be like your work number, right? I mean, how do you handle those different phone numbers, Max? I mean, do you think the idea is that this just becomes, your one phone number over time and Beside is your phone app, your messages app, your MVNO carrier, is that kind of the vision? - The people who adopted already had adopted their phone number as their business number. So their life is kind of like spread across, you know, business context and personal context in the product. They do like to see that they can have multiple phone numbers in the product. So they will actually, actually separate the two within the product or just adopt the Beside number keep theirs at the OS level and play with that. The closest competition, when you think about it, is Meta and WhatsApp for business, because in many ways, this is what Facebook or Meta is pushing, actually, with WhatsApp for business. And those interactions are there. So when I think about the big tech players, it's more than it is Google, more than it is Apple. The real competition is actually WhatsApp for business, because the adoption of WhatsApp for business is huge, and they're really good at bringing AI into the product, and it's quite used. The thing there is people are still searching for businesses on Google, and they're still using phone numbers. Google is not pushing for WhatsApp numbers there. So you still have 50% of business communication happening over the phone and not on WhatsApp. But at the product thinking level, WhatsApp for business is the closest that serve the same type of needs that our customers have. I'm curious about the note-taking aspect of this. That was the first thing that kind of came up when I downloaded your Mac app. I've got three apps in my dock that will do AI note-taking for meetings, Notion, Granola, ChatGPT, and then Google Meet will do it. I was in a meeting the other day, a Zoom, where there were three. There were three individual AI note-taking apps in the call from three different providers. It's super crowded. Is the way that you get into this and you have defensibility because you're actually replacing the phone number? And so if someone's going to buy into that concept, they're going to use you for note-taking because you're playing that kind of game with the context that you have that no one else has? I would see it more from the motion that we have on the go-to-market and who we have product market fit with today. So we're actually on desktop, not even interested in bringing our crowd. And the crowd that we're bringing never heard about Notion, never heard about Granola. They have no idea that these things even exist. They've never seen a call recorder before, though sometimes, you know, half of their calls happen on Google Meet because they're a real estate agent and sometimes they have to show slides and they have. They have to go through a deck and they want the context of their conversation, same as what they had on the phone to live inside the same product. Because if it's scattered among many tools, not everything is connected to their CRM. Not everything is updating the tools around the context is lost for their team. So the lock-in is the context. And then the defensibility is just that we're going after a type of audience that is absolutely not the one of the other winners. So when you look at our product from the tech lens, it's another competitor to Granola, another competitor to Notion Recorder. But if you look at it from the notion of who is actually adopting that product on macOS, the audience is actually completely different. Just say, Alex, it's not for you. It's not for you, Alex. You're not a member of the real economy. You just sit there in a puffy vest and tweet and write on Substack. I just use Granola. It's not for you, sorry buddy. Which is great. They've done like, Granola is an amazing product. Well, so another technical thing that a lot of these services have in common that you'll really need to succeed at is memory. I feel like it's kind of fake when companies say today that they understand everything that you've ever said and do with the context windows being where they are still inherently limited. What are you seeing with that and how are you trying to build better memory systems? Yeah, so that's a use case that we discovered. It's not a use case that is not used by our users. They don't prompt their brain or they don't prompt their context as much as we would, because we're probably using it for creative outputs or to write something or to communicate. For them, the accuracy that really matters is the cold summary. It is the ability to take the right action. In their CRM, sharing the context with their team, and they have very little use case for the prompting the AI. Or what they want to do is like edit their settings or like share something to someone or us doing a call for them, but less so asking specific questions about the memory. I do think it is something we need to solve because as we expand, I don't think we're going to stay limited to the real economy. There is more. We want to do, uh, and we, we want to go beyond that, uh, to the fake economy, our fake economy, uh, I want to break into our tech bubble. Uh, I don't, you know, it is, it is part of the, it is part of the American dream to get there. Uh, but, but, but I think the way for us and the way we've, we've thought about it is that many of the note takers are basically. Storage of notes and information, and they are made to be prompted because that's the only value you can get out of them today. Um, we only wonder based on that context, what are the actions we can automatically take for you? Uh, how do we make that context very much alive, uh, in your experience, in your operating system? If we're going to meet again, uh, in a week, uh, I want to have a, a, a clear meeting prep note based on our past conversation. If I committed to sharing a doc with you, I want this doc to have been shared automatically by the AI. And I think the life context of conversation, when you, when you start thinking about it, not in terms of like storage of notes, but what you can actually take in terms of action on behalf of the users from the context you've captured. And what are the things that you can start automating in these workflows and the value you can deliver? I think you start going away from, uh, you know, uh, uh, uh, an only. Uh, notes and prompting the notes well to the missing context to feel closer to AGI. What about on the customer side though, when they're asking, uh, priest power user questions about the Bible? I mean, if you want to put the whole Bible in there, why not? But can your product search for it? I mean, I feel like that's something I'm still not seeing like with any AI products reliably is actually being able to search for information. Um. In a way that works. Yeah. So I think when they input some of these things, uh, it knows that they exist, but then it relies on the, on the model itself, having in memory, all of that, we don't have a tool calling that would search the web, for example, uh, on the AI receptionist. That is not something, uh, that we've, that we've built. I'm not sure we'll ever build it or that will actually be a clear need. So when people go. Off, what's in the knowledge of the AI receptionist, I think it just falls back to the, to the model, uh, knowledge and the, the model training data. Yeah. I feel like this use case in particular is just so helpful, uh, for folks like the ones that you've heroed, um, people who are not as good note takers as reporters like Alex and I, and just need to remember all this stuff. Um, but I mean, is there some limit? Is there a limit to like how far back you can go memory wise with like your calls, asking things from the past, what are some insights that all my customers are telling me things like that, that really rely on like a huge data set to help move the business forward, or is that not really on the table yet? No, I think like that, that's the evolution from where we're at to, to, to where we're going. And there, I think until it's fully ready. That may be thousands and thousands of calls, right? Yeah. Well, there, I think you, you enter more custom prompting rag building, and you're actually probably building the product around the limitation of the, uh, of the memory of the model. And as you build that, or as you invest there, you know, it's just going to be sold six months, 12 months from now from the, from the model itself. So that's the way we approach it, that there are some things. Well, we are wondering if we should invest time into building because they might just self, uh, solve six months itself. Yeah. How do you think about that? Isn't it? Yeah. A lot of people are going through that right now. I'm thinking about, you know, that, and how do you build a company for AGI if that's going to happen or just the models being able to replicate your company potentially? Yeah. Um, one, I'm excited, I think, uh, cause what? If, if we get there, what it means is absolutely, uh, amazing and the technological progress we would have made, uh, I, I am very grateful and lucky to be alive at that time and in the industry, uh, on the field, because that's the most exciting of times in my opinion. Um, but beyond that, the way the importance of unique context that you need to capture. So. There are contexts that you're going to be able to capture through an API, there is context that you'd be able to capture because you You can build a Mac OS recorder like everyone is doing. But then there's going to be unique, highly valuable context. The phone context is particularly valuable because what you need to do to capture a phone conversation is extremely hard. To me, the only defensibility as you go is going to be, for us, the consumer-grade experience because if you don't have that DNA built into the team and if you are coming from the B2B world or the SaaS world, it's going to be extremely hard to rewire your brain to consumer product design, product building, and product distribution. The other part is finding the right data that other players are not going to easily capture, which for us is the phone. One of the questions we'd like to ask is, if you're thinking about the future, is what do you optimize for? I know you can look at any number of things from call length, call volume, how many questions people are asking, their conversion with customers, how many callbacks. A conversion to a religion. Yeah, conversion to a religion. You could be the first. What are you optimizing for? Accuracy of actions we can take. Did your customer get a great, interaction with the voicemail receptionist? Were we able to properly book in your calendar the call that they ask? When you come back to the product, you read a past conversation, does the summary look accurate? So the value of the product becomes extremely passive as we go. So what really matters is that we're able to measure the accuracy of the action we take on your behalf as a user and get as close to 100% as possible and as many as we can. Yeah, speaking of those actions, I mean, it just makes me think a lot of those actions were for many people taken by humans in the past, whether it's, you know, a scheduler, whether it's at a hair salon or a car dealership or otherwise serving the real economy, growing up in the real economy. I think you know where I'm going with this. What's your latest thinking about, you know, the jobs being lost, replaced, changing for these people? What we've seen so far is that, we were more coming as a help to people who are already very busy and could not actually delegate to someone. I don't think we've entered a place where we've replaced anyone. What I see is we deliver value for a lot of people where they can actually offer better human-to-human experience. If you've ever been to this super high luxury hotel, where they have concierge, you give your name once, you give your preferences once, and that's it. You won't have to repeat yourself and you will just like flow through your experience. I think that's the beauty of what we can offer. People feel more cared for, the interactions become more meaningful, and there's really beauty in that because as a human, you can't actually do that if you don't have some machine help. So one more question for you, Max, before we let you go. Who do you think has the best taste in tech right now? It's a good question. And don't say Johnny Ive boat lamp. No. It could be Jack Dorsey, to be honest. Like, I really love his latest. And then there's Antoine. You know Antoine Martin from Amo. Like, he's probably, you know, the taste goat of consumer. Well, we don't talk about Jack nearly enough. What's Jack doing right? That you see? He's very, he's still focused on Square. He's also big on Bitcoin at the moment. He also built a messaging app that works with Bluetooth devices. And I really enjoy how he always comes down to like, at some point, he does all these big things and just like comes back and code himself a little app and like start a movement with it. I just find it really cool that he just gets down to this like low level in the arena. And just restart from scratch and seems to be excited about it. Yeah, I find this to be inspiring. Some people think he's Satoshi. Have you heard that, Max? That he's Satoshi? No fucking way. Yeah. No way. That's a rumor. No way. All right, well, let's leave it there, Max. Thank you so much for coming on Access. We'll chat soon. Chat soon. Nice to meet you, Max. Thank you, guys. Thanks to Max for coming on this show. Oh my God, that's what I was going to do. We have a new show, so please follow, like, subscribe everywhere you get the podcasts. You can also find us in video via AccessPod on YouTube. If you like this episode, please subscribe, leave a five-star review, a great review, a glowing review, a thumbs up. Share it with a friend, even better. And you can find me @hamburger on Twitter and at meaning.company. He did give us permission to do this, right, Alex? I think he did. You can find me @sources.news in my newsletter on Substack. Access is part of the Vox Media Podcast Network, and this show is produced by Hooked Creators. Bye. Au revoir! Booking.com is the easiest way from a day surrounded by noise-- to a state surrounded by nature. That's nice. Go on, book it. It's easy. Booking.com. Booking.yeah. Fall is the perfect time to refresh and reorganize your space. At the Home Depot, find power tools and tool sets starting at $50 to help tackle DIY projects. And if you're looking for a new home, you can find a new home in the home depot. Go on, book it. It's easy. Booking.com. Booking.yeah. At the Home Depot, find power tools and tool sets starting at $50 to help tackle DIY projects, home updates, and more. Whether you're drilling brackets to support new shelving or sharpening your hedge trimmer blade with an angle grinder, the Home Depot has the tools you need to check projects off your list. Shop Labor Day Savings at the Home Depot and gear up for fall projects with the right tools to keep your projects moving. you