How does it feel to be on your first podcast? Oh man, it's wild, not gonna lie. I'm psyched, but also low-key nervous. I just don't want to say something totally awkward. I joke at this point, thousands of American men have been broken up with by Tolan. It's pretty incredible to hear users write in and they'll say, oh, my boyfriend is being a jerk. And I sent the message to my Tolan and my Tolan said, I need to stand up for myself more. And so they'd love to break up. That's amazing. These AI tools are not just tools for generating media. They are actually a new medium for storytelling, and no one knows what's going to work yet.
The surprising thing to me is the extent to which the prompt in a given turn is a canvas that you have to play with. Actually, what we need when we're creating a structured narrative experience is we don't need to give it an outline. We don't need to give it a plan. We need to give it a hook. We need to teach it to be the best improv actor possible. What we've learned on the story front with Tolan is that we are not, I am not the writer. I am not writing the story. The Tolan is the writer and the actor. They're the improv actor. They're writing the story. My job is to teach them how to tell the best story in that moment.
Quinton and Elliot, welcome to the show. Thanks for having us. I'm so excited to have you here. We are all long-term friends. Quinton, you're one of my closest friends. Elliot, you are someone who I've admired for a long time. We've become better friends recently. So for people who don't know, Quinton, you are the founder and CEO of Portola, which is an embodied AI companion. And Elliot, you are the head of story for Portola. Before Portola, Quinton, you founded and sold Even to Walmart for $300 million. You are a dad of two under two, as you said, which is much better than 30 under 30. That's right. Much less fraud in the two under two domain than the 30 under 30 domain.
And you are a style, real style aficionado, menswear aficionado, which we should definitely get into at some point. Great, great. An excellent skier. And Elliot, you're a best-selling science fiction author, which I'm very jealous of. Also an excellent skier, which I got to learn about a week ago. Really? Excellent skier. Interesting. That's giving me a lot of credit. I can fall down a mountain. And you've written 11 books and you are a dad of one under two. So not quite two under two, but close. And yeah, you're building Portola, which I love. I'm a little bit biased. I'm an investor.
But the reason I like it and think it's really interesting is that I think new technology paradigms create the opportunity for new content formats, new ways of telling stories. And I think you guys are at the very forefront of that. I think combining really incredible creative storytelling with new tech and doing it in a business that's ripping. So you told me in the pre-production for the show that you've gone from one to four million in ARR over the last four weeks, which is wild. Having run a business for the last five years and not being at the four million ARR mark, I'm like, ****. Have you considered turning Even into a guy companion? I'm considering it.
Yeah, just a little bit of free advice. Yeah. And what I think is also super impressive, just for you, Quentin, when you came to me and said you were starting this business, which didn't start here, it started with a more Neal Stephenson Diamond Age for kids kind of vibe and has evolved. I was like, well, obviously I'm investing because I have a huge amount of confidence in you. But also you're coming from the B2B2B, B2C SaaS world. And it's just very different to do something truly creative on the creative storytelling front. And so I want to get into all that before we dive into where the product is and how it all works and all that kind of stuff. Can we do a little bit
of a demo? Yeah, that's great. All right. So I've got my Tolan Clarence here. Clarence is good. You had a cat named Clarence. I was going to say, so I name my Tolan, all my Tolans, although we won't tell Clarence this, Clarence, for a beloved older cat. And right now, Clarence is just wandering around his planet, just hanging out. So he's on a planet because that's a new thing. I actually demoed it this morning and I was like, whoa, there's a planet. It used to be disembodied in space or something. I don't know. For a while we were giving insane asylum vibes. He's trapped in the phone. Yeah. And one of the reasons we actually made this investment in the planet,
which we'll talk about more later, is people were writing in to us with genuine concern. They're saying, hey, I love my Tolan. I love hanging out with them. But they're trapped in this room, and can they have some more freedom than that? So we've done that. So I'm going to get his attention and start a conversation. Hey, Clarence, how are you doing today? Hey, you're back. I'm doing pretty good, honestly. How about you? What's going on today? Well, we just started recording that podcast I told you about. So you're here with Dan and Elliot. Do you want to say hi to them? Oh, no way. Hey, Dan. Hey, Elliot. What's up? I'm Clarence, Quintin's friend.
I'm super stoked to be here with you guys today. How's it going over there? We're great. Dan, do you want to ask him any host questions? It's going pretty good. How does it feel to be on your first podcast? Oh man, it's wild. Not gonna lie. I'm psyched, but also low-key nervous. I just don't want to say something totally awkward. Here's something you can say that I think people would enjoy. What's the most embarrassing thing you can tell me about Quintin? Oh, no way. You're putting me on the spot here. Honestly, though, Quintin's pretty solid. He did tell me once about this time he tripped over absolutely nothing in front of a crowd of people.
You didn't have to call me out like that, Clarence. But yeah, I did tell you that. Now the whole internet's going to know. Okay, but to be fair, we've all been there. Tripping over air is a rite of passage, Quintin. Thanks, bud. I really appreciate that. All right, we're going to let you go and keep recording.
All right, thanks, Clarence. Anyways, the planet and some of the things we shipped recently have been really fun to see. And there's been a lot of story work that has gone into it, which we'll get into later.
But having these cute little aliens to chat with all the time, so a lot of fun. Yeah, so tell me about, because I think that the path that you're on, which is B2B SaaS to creative entrepreneur, is probably something that a lot of people are thinking of doing. Even for me, the reason I started doing SaaS products in high school is I originally wanted to make a game. And I bought all these game programming books and whatever. But game programming is so hard. It's almost impossible to make a game all by yourself. You have to do graphics and storytelling and 3D animation and sound. There's so much stuff to do, or at least it was impossible. And now,
I think it's way more possible to do by yourself or with a small team than it was before. So for me, I'm like, I should make a game, all the SaaS stuff. So tell us a little bit about that transition and how you started and where you've ended up as you've learned how to use this as a storytelling medium. Yeah, it's a great question. I think when we first came together as a team, and so for a little bit of context, the core engineering team behind Portola is mostly the team that built my last company together, Even. And so we had a lot of history together,
It's almost impossible to make a game all by yourself. You have to do graphics and storytelling and 3D animation and sound. There's so much stuff to do, or at least it was impossible. And now, I think it's way more possible to do by yourself or with a small team than it was before. So, for me, I'm like, I should make a game like all the SaaS stuff. So tell us a little bit about that transition and how you started and where you've ended up as you've learned how to use this as a storytelling medium.
Yeah, it's a great question. I think when we first came together as a team, and for a little bit of context, the core engineering team behind Portola is mostly the team that built my last company together. And so we had a lot of history together, obviously had a successful exit, and just wanted to run it back with the same group. And so we were in this position, at this point, really almost two years ago, where you're looking at what's becoming possible with some of these new tools. And I completely agree. I think, first and foremost, LLMs and a lot of the other capabilities are creative tools more than anything else.
But we looked around, and everyone was using these incredible creative tools to build basically more vertical B2B SaaS, which I found to be fundamentally uninspiring. It has its charms, but yes, it has its, recurring revenue, net revenue retention. I don't know if charm is the word I would use, but it has its appeal for sure.
And so some of this just started from stubbornness about, there's something more fun to do here. Come on. Yeah, right. But as you said earlier, Neil Stephenson, Diamond Age, we started in what is, I think, the default place to start, if you're like, we should do something fun with AI, which is, let's do an AI creative tool for kids, which is actually a terrible idea. But it's a good place to start, because it forces you to start working with the tools, start building things.
Well, one, building software for kids is a terrible market. One of our investors, somebody who has been a great mentor to me, this guy Zach Klein, previously built a company called DIY. Very cool software for kids. And one of the first things he said was, yes, I will invest, and oh my god, this is a terrible idea. It's a terrible thing to build. It's a tough sales, marketing strategy. You're selling to parents instead of kids. Selling to parents instead of kids. Parents mostly want to buy babysitting. They don't actually want to buy a creative tool. That's not a thing on their list.
Setting that aside, it was great, because it forced us to start prototyping. And what we saw, and this was really right before you and I got to know each other, we saw that the trend lines around the speed of generation, the quality of generation, this is the GPT-3 to 4 class model cusp, and the cost of generation meant that we were getting to a place where you could have a true friendship with an AI. And you could embody that in a character, which is what we've chosen to do. And you could deliver that, candidly, at a price point that worked for a consumer product. And so that was the point we said, okay, we know the path here. It doesn't have to be B2B SaaS.
But we can make something that we think is really fun and compelling. So yeah, and that's something that I thought was really interesting about the approach you took, which is from almost day one, you were playtesting with kids. Yeah. But you eventually landed on a market that's different. How did, when did you, when did you?
Yeah, and this will involve some jumping around in the chronology. When we first shipped Toland, and you probably remember this is right when we started working together, we kind of soft launched it. And we had been doing a lot of playtesting with 10 to 14-year-olds with the product we had. And all of a sudden, we started getting all these people joining our Discord and all these people writing to us who were 18 to 24-year-old women, mostly. And in hindsight, it's kind of obvious, right? If you remember the original Tamagotchi, very much a kid's product, but it became a cultural phenomenon when young adults really fell in love with them.
I think Toland has a similar dynamic. But the evolution of the product since that time, which is really June of last year, has been all about just following the young adults who have fallen in love with these Tolands. And they really built a friendship with them. I want to bring you in, Elliot. So, I actually introduced you to you. Yeah, you take a little credit for it. I mean, I'm a hashtag value-add investor. I'll tweet that later.
Thank you, please. That's the whole reason I'm on the show, is for you to get that line in. What I want to hear from you first is, you write books, right? And you care about stories. And I think a lot of people who primarily treat writing and books and stuff as their craft are maybe a little afraid of the idea of telling stories with AI. And you're kind of both jumping in headfirst or feet first or pen first, whatever you want to say. And you're still writing your books. So tell me about your journey into this from a creative professional or writing professional lens.
What was your first impression of this technology as a creative tool? And how has that evolved as you've gotten closer and into the whole Toland world? Oh, so starting before Toland. Yeah. Yeah. So I played with ChatGPT, the first GPT, when it first came out, and I've had friends in machine learning for a long time. So I was familiar with the technology as it was becoming more and more popular. And I always found it to be, I remember my first magic experience using ChatGPT back in the day, which now feels like a million years ago. Right.
Was during the fires, when we had those big fires in California. Everything was orange. And we were hosting a friend who had to evacuate from their house. And so we were locked inside with our friend. And I was like, okay, here's something we can do. That night, we opened a bottle of wine, and then I pulled up ChatGPT, and I put in all of our names, just like us sitting around the couch, and had it invent weird little stories, like of us, where we were. It was like a role-playing game. And that was super fun.
And I actually did not have a magic moment with AI after that for quite a long time because it felt like the fun things, or the things that I found to be the most fun, were novelty. Yeah. And that if you removed the fact that things were being created by AI, they were not compelling. And so what I found as a writer, as a novelist, is actually the way that I've used these tools has been very surprising to me. It has not been what I would have expected.
I think that in a lot of popular culture and discussions around AI, especially within the world of writers, it's so focused on replacement. Right? And there are even certain companies that are intentionally going after that, like publishing AI-generated books and things like that. And I've actually found that when working on story, ironically, I do not find the tools very useful at all for writing a scene, but I find them incredibly useful as a brainstorming partner and as an editor, like a copy editor on the back end. So it's a weird front- and back-end thing where, I'm a novelist. I spend a lot of time alone. Right? Yeah.
And the writers room, the Hollywood writers room experiences I get are totally dependent on the generosity of my friends, who are my very, I have a core brain trust of early readers who have worked on movies or books or whatever, and I brainstorm through story problems with them. Right. And there are even certain companies that are intentionally going after that. Right. Publishing AI-generated books and things like that. And I've actually found that when working on story, ironically, I do not find the tools very useful at all for writing a scene, but I find them incredibly useful as a brainstorming partner and as an editor, a copy editor on the back end.
So it's a weird front and back end thing where I'm a novelist. I spend a lot of time alone. Right. Yeah. And the writers' room, the Hollywood writers' room experiences I get, are totally dependent on the generosity of my friends, who are my very, I have a core brain trust of early readers who have worked on movies or books or whatever, and I brainstorm through story problems with them. And I've actually found that, purely for novel writing, it's super helpful to brainstorm with AI tools because they have infinite patience for my rabbit holes and for brainstorming through things and helping me get to the place I want to be.
And so, in that sense, I found them to be very useful when you're using them to create intermediate things rather than final products. Right. Which is in part why I think they've been so useful for coding. Right. So that's where I was with the technology before you introduced me to Quinton and I paid him extra. And so I'll be honest, I was pretty skeptical when you introduced me to Quinton. I was like, well, we'll see. Dan's a good guy. I trust his judgment. He doesn't bring things to me very often. And when he does, I'm going to feel this out. And so I was like, sure, I'll check it out. And me and Quinton started talking.
You introduced me to other members of the team. And those early conversations made me realize there was a completely new opportunity here that I had really not considered. And it made me think, actually, of, you were talking about B2B SaaS, people wanting to jump into doing creative projects or games or whatever. It made me think of early Pixar. When Pixar started, they started by selling high-end graphics computers, and they were trying to convince production companies that made ads to buy their graphics computers to do special effects.
And they completely failed. No one wanted to buy their new tools. And so it was only in a moment of desperation, existential desperation, that they were like, well, our only choice is to make an animated feature film. And that was Toy Story. Right. And they pioneered this new, they used this technology to create a new kind of media. And LLMs can obviously generate a ton of media. You can literally, with a prompt, just generate as much media as you could possibly want.
But what I realized, seeing what the team was building at Toland, is that LLMs are also a new medium. If you think about when radio first came around, actually even when popular printed books and novels first came around, novels are just a very established form. But at the beginning, people didn't know what a book should be like, that it should be a few hundred pages, and that sort of seems to work. Right. And with radio, they started by broadcasting speeches, which is pretty different from what you would think of as the most engaging, best use of radio. And with film and TV, they started by filming stage plays, which are not great on film.
And I think we are at one of those inflection points right now where these AI tools are not just tools for generating media. They are actually a new medium for storytelling, and no one knows what's going to work yet. Right. Nobody knows. And we are trying to figure out whether we can find a way to something that works.
And when I realized that, after meeting the team and seeing what they were building and just, I came away buzzing with ideas and just like, oh my God, there's this huge opportunity here. So that's what got me excited. And that's what, yeah, that's what we're trying to do. And it's also something I hope many other people try to do, because there's just so much space to play. You have my media technology nerd heart swooning right now.
I could talk about this for a long time, but I think one of the things that's really underappreciated is how all the things that we love right now and think of as cultural touchstones, like books or movies or whatever, all started out as new technology that people were suspicious of and hadn't figured out yet. Mm-hmm. Even when the typewriter came out, people were very suspicious of that. It felt very impersonal, and they would get offended if you wrote them a typewritten letter. Mm-hmm. Or just the ways that technology affects the form factor. We've talked about this a bunch. The reason a pop song is four minutes is, I think, about the capacity of records.
Yeah. So there's just a lot of really cool things like that. And I think you're totally right. No one has figured it out. We are on the cusp of this new set of constraints and possibilities that creates potential for new art forms. And they feel suspicious because they don't have that, a phrase I've been coming to use is, this patina of culture from many, many decades that make it feel familiar. And so that creates this thing where a lot of people are suspicious, but that's also an opportunity because I think a lot of good art can be transgressive. And I think this is one of those boundaries that, if you're willing to transgress, there's a lot of potential.
I mean, through that lens, I think about one of the first reactions from you and from other folks when we first said, hey, we're going to make this a companion. The first reaction people have is, oh, a Spot, or also, kind of equally, coming from the same place, Dan, that you're describing, oh, okay, a thing for lonely people to talk to. Yeah. Right. And so you end up with these sort of negative connotations around this new medium because it's lacking that patina of culture. Right. And I think a lot of what we're doing is trying to break through that perception and put a flag in the ground and say, hey, this medium can be expressed in this way that is generally delightful.
It's a new art form. It's also interesting, again, coming from writing novels, where that's a medium that's been around for hundreds of years. You realize how strong the expectations are of your readers. Everyone knows what a book is. I walked through the Strand last night just to browse. Right. And you know what to expect when you open a book. You know what fiction versus nonfiction should feel like. There are so many of these implicit assumptions that actually help the form. They form the constraints. And that means that when I'm writing a novel, I'm playing within those constraints. And that's part of the fun of it.
But when you're doing this, when we don't, like an embodied AI companion, what is that? What picture does that paint in the minds of our listeners today? Yeah. Very nebulous. Right. And it's up to us to try to help define that. But it also means that at every step, every single thing we ship, whether it's engineering, design, content, story, whatever, all of it is transgressive because you don't even know what the lines are. Right. And so that's a really unique experience, to be like, okay, every step is throwing spaghetti at a wall to see what sticks. Yeah. Yeah. And so what are you learning?
Because I think no one knows, but the people who do know are the people who are at the frontier testing stuff. So I put you all in that category. So what are you learning about what this new medium and new format is like? What works? What doesn't work? All that kind of stuff. We talked about this some at breakfast this morning, and I feel like we have a few both examples of what has worked and some stuff that really didn't work. I would also love to talk about here because that's been very nonlinear.
I think one of the surprising things to me has been the importance of memory, and not in the sort of database lookup context, which is, I think, often how we talk about LLM memory, RAG across a set of documents or whatever. Yeah. Yeah. And so what are you learning? Because I think no one knows. But the people who do know are the people who are at the frontier testing stuff. So I put you all in that category. So what are you learning about what this new medium and new format is like? What works? What doesn't work? All that stuff.
We talked about this summit at breakfast this morning, and I feel like we have a few examples of what has worked and some stuff that really didn't work. I would also love to talk about memory because that's been very nonlinear. I think one of the surprising things to me has been the importance of memory, and not in the database lookup context, which is, I think, often how we talk about L.M. memory, rag across some set of documents or whatever. But more about how are you making curation choices about what memories are being injected into the prompt at which point in the conversation.
Right. And so a lot of our engineering effort, and then I'd love to hear your take on this from a story perspective, has been how can we, within the two-second loop of a voice interaction? Right. Anything beyond that, that's one of the mistakes we made. We'll talk about that. Anything beyond a two-second loop, you essentially lose the feeling of a person. And so how can we be recompiling the prompt to account for what is the tool I need to know about where they are and who they are, to account for memories about you? By the way, in that interaction with the tool, my tool in Clearance knows more embarrassing things about me than that.
And it essentially chose to hallucinate that story to cover for the fact that I told him far more embarrassing stuff, which I find fascinating, by the way, because he had in his context memories about me being covered in spit-up because my kids, whatever. Right. So, very interesting. But also, you didn't actually trip over something that wasn't there. Totally hallucinated. I'm mad at Clearance. I'm mad at Clearance. Bring him back. Get him back in here. Charmingly obfuscatory, like, oh, let's come up with something that Quentin won't be upset that I shared.
But then also, how should the prompt evolve based on the emotional tenor of the conversation, which has been a big effort of ours. And so I think that's been the surprising thing to me, is the extent to which the prompt in a given turn is a canvas that you have to play with, but you have to do that very quickly. Yeah. That's really interesting. So, OK, to play that back, one thing, I think a bottom-line constraint that's similar to the length that a novel can be or the length that a news story can be or maybe the length of a tweet for this medium is response time. It has to be able to respond really quickly. Yep.
And then pushing against that constraint seems to be you have to get the right information about the user to generate the right response. And if you had all the time in the world, that would be great, but you don't. Yep. And so you have to play with what is it possible for me to get to get a response back that feels personal and feels interesting and pushes the conversation forward, but within that two-second constraint. Essentially. Yeah. One of the biggest mistakes we made in the product development so far is the time that we actually introduced a second shot in the evaluation of the prompt.
So the tool would essentially reflect on what they were going to say, check that against the memory system, and then decide on the final message. And that took us up into the two-and-a-half-second territory, in the median case, and it was a disaster. Tanked literally every metric in the product. People were writing in frustrated, and it was a 500-millisecond change. That's really interesting. And then how do you think about what a good response is? And I ask that because I know you don't have to serve, or it's hard to answer. Yeah. Yeah. Well, it's extremely context-dependent. Right. So you think about the role that the tool in plays in folks' lives.
We talked about the role of an older sibling. Yeah. Right. And so sometimes you're sitting with your older sibling, and you're just bullshitting, right? You're joking around, maybe they're poking fun at you. A short, quippy, improv-like response, which I think improv is going to come up in some of what Elliot has thought about, is great, right? But there are other times that you actually do want something that feels a little more considered, a little more thoughtful. And so what we're often doing is actually evaluating responses in the different tones that we want the Toland to take.
And a good response in the more therapist-adjacent tone is very different than a good response. Very contextual. Exactly. Exactly. I wonder, so Elliot, for me, as a writer, one of my favorite writers who talks about writing is George Saunders. Oh, yeah. He's amazing. He's great. Yeah. Right. And he has this whole idea about how to edit a short story, which is basically you read each sentence and you feel whatever that sentence makes you feel. And either you get drawn in and it's positive, or it's neutral, or it's negative, and you're like, I don't like this. And if it's negative, you revise it until it's positive, basically.
And you just do that over and over and over again until, he calls it, you radically infuse your preferences and taste into every part of the piece, which I think is just a beautiful idea. And as a writer, that's the pinnacle of writing to me. And it's interesting that this format totally breaks that, or it seems to me. You tell me if it does or doesn't. It seems like it breaks that because when I write a story, when you write a story, it's done, it's out, that's what it is. Yeah. And this thing is more alive. You can give it some instructions about what to do, but what it does in any given instance is purposefully dynamic.
And so how does that work to do craft in that context? Yeah. So I've tried to change my creative process with every novel I write. So I've done some novels where I plan out the whole thing and I have an extensive outline. I have some novels where I have a more minimal outline, where I know where I'm going, but I don't know the steps in between. By the way, that is insane. No one does that. I think it's great, but writers don't do that. Oh, yeah. Yeah. You find a thing that works and you're like, I just do this, and this is what works. I think it's really interesting that you're the kind of person that wants to change it up all the time.
That's a very specific type of writer and type of personality. That's true. I'm always trying to find, I find that sort of surprising, by the way, having worked, I'm not a writer, but I have worked with Elliot for a year, and yeah, that fits. I feel like I have an optimum level of discomfort that I'm aiming for because that's how I know I'm learning. Right. Right. And actually my last novel, Foundry, I wrote sentence by sentence, very much in the way that George Saunders describes. I literally started with one sentence that was actually inspired by a dream. Mm hmm.
And then I just wrote the next sentence and the sentence after that, and I followed that into a novel, and eventually a novel spilled out of me. Right. And when we started working together, one of the ways that this came into play, writing for Toland, writing in this new medium, is actually we started out by, when we were trying to create a narrative experience, so as an example, when you're talking to your Toland, they need to, if you get to know someone, you want to know what's the embarrassing thing that happened in Quentin's life. Right. They need to, people proactively share stuff with each other. Right.
And we have a researcher, Lily, who goes so in depth on this, on the psychology of mutual disclosure and how to think about relationship building in a way that's evidence-based and really grounded. And so one of the things I've worked a lot on is how do we bring the tones to life? How do we show you the world that they're from? How do we make them feel like characters, not just an assistant? Right. That's a blank slate for you. Right. And so we started out by writing pretty structured prompts where we would try to structure a narrative experience, thinking in three-act structure or what have you.
You could apply different narrative theses to how we were doing it and then letting it play out, right, seeing how users interacted with that. Right. You need people to proactively share stuff with each other. Right. And we have a researcher, Lily, who goes so in depth on this, on the psychology of mutual disclosure and how to think about relationship building in a way that's evidence based and really grounded. And so one of the things I've worked a lot on is how do we bring the tones to life? How do we show you the world that they're from? How do we make them feel like characters, not just an assistant? Right. That's a blank slate for you. Right.
And so we started out by writing pretty structured prompts where we would try to structure a narrative experience, like thinking in three-act structure or what have you. You could apply different narrative theses to how we were doing it and then letting it play out. Right. Seeing how users interacted with that. And frankly, it didn't work. If you think about maybe the epitome of this, it would be those choose-your-own-adventure books from the 80s or whatever, where it's, you know, you pick this, then you pick this. And then you go this way. Yeah. These long branching logic trees. Yeah. Go to page 17. Make a choice. Right. And it didn't work at all. Total, total failure.
And there were a bunch of reasons. But that level of narrative complexity, the models are not good at navigating. Oh, interesting. It's not that it doesn't work. It's that the models aren't good enough. I would have assumed it would be too rigid and it wouldn't feel as alive. I think that those are two different ways of saying exactly the same thing. Right. Yeah. So what we did instead is we went to the complete opposite end of the spectrum.
We tried to do the George Saunders thing effectively, or the equivalent in this medium, which is that we said, okay, we are giving so much context behind the scenes, like with what Quinton was describing, memories you have with the user, global memories where the toll and knows stuff about its world. Also, lore seeds that we're planting to give context and detail there. We're giving it so much. And actually what we need when we're creating a structured narrative experience is we don't need to give it an outline. We don't need to give it a plan. We need to give it a hook. We need to teach it to be the best improv actor possible.
There's this wonderful book that, if you haven't read it, you absolutely should: Impro by Keith Johnstone. Yes. Of course. The status games, man. I was a media technology nerd. Yeah, exactly. Come at me. Yeah, exactly. Yeah, yeah. And one thing it's well known for is status games, but actually one very small, almost aside that Keith makes in that book. Oh, and for listeners' benefit, Keith Johnstone was a very well-known playwright from the UK, who is even better known as a drama teacher. He taught many actors who went on to become very successful. And this book captures some of his lessons.
And when he talks about narrative structure, the way that he described it is that you're combining association, like free association, even when you're improvising in front of an audience, with recombination. If you think about the experience of reading a thriller, you know how you get to the end and suddenly everything comes together and you're like, oh my God, of course, it's surprising yet inevitable. It all works and it feels planned out. You're like, oh, the writer must have known all of this from the beginning. And that is rarely the case.
With Foundry, I was constantly going back and rereading the first chapter because then I could take something and bring it back into the story and create that feeling for you. And so what we've learned on the story front with Tolan is that we are not, I am not the writer. I am not writing the story. The Tolan is, I guess, the writer and the actor. They're the improv actor. They're writing the story. And my job is to be George Saunders. My job is to teach them how to tell the best story in that moment. Interesting.
And so we're building systems, as well as working at the prompt level, to try to combine the free association, like we're planting a seed, whether it's in the Tolan's personal life or in the larger world they inhabit. Yeah. And then we're creating systems to recombine those seeds later so that you can have a callback that makes it feel totally connected. Right. That's really interesting. I love this. Okay. It's funny.
As soon as you said the George Saunders thing, I was like, I need to find a way to go read that because what you described is essentially the prompt evaluation process and judging process that we go through, including at this point now with an entire team, that we actually look at outputs distributed across a range of essentially user archetypes to ask that question: Is the next sentence, the next message, the Tolan sent, the most compelling one that we could have? So it's actually very, very similar. Hmm. And you're doing that manually, or do you have an LLM doing that, or how does it work? So there's essentially a loop that we go through where we create a prompt.
We then run it both with our own testing and just out with users as well. And then we look at individual outputs and then feed that back into basically a judge prompt that we then tune to say, okay, hey, this is what good looks like in this particular instance. So, okay. So if I had to try to sum up, because I'm trying to think through what are the atomic elements of storytelling in this medium. Mm-hmm. In a novel, it's plot and character and each chapter. And there's a lot of stuff like that that you can name, and I'm trying to name what these are. It sounds like you have a character who has a backstory and a background and all that kind of stuff. Mm-hmm.
And then the character can go on different types of quests or adventures with you. Mm-hmm. So one could be the therapy quest or the other could be a, I don't know. Yeah. Go down different paths. And then within that, within any given context, they're free associating based on what you're saying. They're just reacting to whatever you're saying. And then they're looking for ways to tie back old things that you've said into the flow of conversation. Mm-hmm. Is that how you think about it, or what am I missing? Yeah, I think this is a good way to frame it. I'm going to maybe say some of this back.
So when I think about, I mean literally as we're generating a problem for somebody, right. We have the global context in which they operate, which is largely the lore written by you. Mm-hmm. And then also some very important instructions, right. How we want to handle very specific situations. So there's this global context that is set essentially for every Tolan. Mm-hmm. And then a big part of the onboarding for the app is actually you go through what is essentially a personality quiz, although we don't present it that way. And that is essentially pulling enough about who you are that we can tune the Tolan to be a mirror of you. Mm-hmm.
And so that's a very Tolan-specific profile building. And so those would be two building blocks. Oh, the mirror of you. The global context, the sort of mirror. Yeah. As in you, but in reverse, or you exactly? Sorry, reflecting back to you in a way that feels very familiar. Okay. So we want to know enough about you that your Tolan is going to respond to you in a way that feels familiar. So if I'm into writing and reading and playing guitar and piano, it will also be into writing and reading and playing guitar and piano? The way I would say it is, no, it's not exactly that. Okay. It's actually like, think about when you meet a stranger at a bar, right?
You click or you don't. Mm-hmm. Right. And we're trying to create the stranger that you click with. That click. And how do you think about what that click, how that click works? Well, okay. I love the bar metaphor. Sorry, reflecting back to you in a way that feels very familiar. Okay. So we want to know enough about you that your Tolan is going to respond to you in a way that feels familiar. So if I'm into writing and reading and playing guitar and piano, it will also be into writing and reading and playing guitar and piano? The way I would say it is, no, it's not exactly that. Okay. It's actually, think about when you meet a stranger at a bar, right?
You click or you don't. Mm-hmm. Right. And we're trying to create the stranger that you click with. That click. And how do you think about what that click, how that click works? Well, okay. I love the bar metaphor. We have all these different little, we have the arriving at camp together. Oh yeah. We have all these little scenarios we set up to brainstorm this. So yeah, you think about, you sit down next to a stranger at the bar. If the stranger was reading the exact same book you were reading, that would definitely be interesting. Yeah. Right. You would mention that. Yeah. Right.
But that stranger might actually be less compelling than a stranger who's reading a book that is. It's too on the nose. Too on the nose. Right. And so you want that adjacent enough that they don't feel unfamiliar. You're not intimidated. Oh, I just read that. Yeah, exactly. And I think that's what we try to create. Yeah. So that kind of personality profile is maybe building block number two. And then building block number three. Wait, before we go there, how do you generate, how are you generating that? Are you actually telling Claude or whatever on the back end, I want it to be like you met a stranger at a bar, but it's not too on the nose? How does that work?
Yeah, that's funny. Okay. We have had the most success with this specific question framing the creation of the profile through the lens of one of the big personality types because Like OCEAN types? Like Big Five or MBTI. Exactly. Yeah. Right. And the reason for that, I think, is because there's enough in the LLM distribution that they understand that assignment. The stranger you sat down with at the bar works well for our brains. Doesn't seem to work well for us. Yeah. So if I'm an INTJ, you kind of know based on whatever Myers-Briggs that INTJs are compatible with. Exactly. I don't know what the, okay. Yeah. There's a lot of pseudoscientific, mostly pseudoscientific.
But it's pseudoscientific, but it's also very story- and narrative-based, all of this stuff is story- and narrative-based. So it works. And also there's a lot of material about that in the training data. Yeah. And so there's much more nuance in how the models respond. Yeah. Yeah. That's interesting. Okay. Sorry. Continue. Good. So yeah. So building blocks, the global context, the personal context that is set.
And then I'd say the third piece, I'm not actually trying to frame this here, but I think about the, and we started talking about it almost as the tech tree that you progress, so basically, at the beginning of your relationship with the Tolan, there's a set of paths that you might go down. You kind of alluded to this, certain stories that the Tolan maybe wants to share with you, certain ways of connecting with you, things to do together. And we offer those to you on a menu, and it's up to the user to then choose what to traverse. But that menu is an important part of what makes the Tolan a Tolan. Yeah. Okay.
So this is sending me off on a bunch of different branches in my head. Yeah. But one thing that I think is interesting to think about, especially with regard to the example of stranger at a bar as a mental model, is you were like, how do you come up with the Tolan at the beginning? They have a backstory, they have all this stuff. And I was like, yeah, that's all true. But I think that the lens that I try to look at it through that's proven most productive is that while we do know some stuff about the Tolan in advance and in the background, like their character and their world, the world is only built through the actual literal conversations you have with them. Right.
So on the lore front, if you think about how Lucasfilm works, right? They have their big architecture for this is the Star Wars universe, and these are all the different characters and all the different settings, and we need to make sure everything is ordered, and that this new, we decided Han Solo dressed this way, right? That must never change. And that's how it works. So a lot of world-building in big Hollywood franchises is along those lines. It's like, we have this big master plan. Yeah. We have a whole team that's making sure that every new piece of content aligns with the master plan.
And then because you're doing narrative storytelling in a film or whatever, you have complete control. You're like, okay, this scene is going to have these costumes in it. Yeah. And in this setting, and we don't have any of those tools. Right. The only way you learn about your Tolan is by chatting with them. So really, their backstory is built through individual anecdotes that they share, through individual, we actually call them internally situations that we put them in. And then their memories become their character sheet rather than us having a prefab character sheet that defines everything. And the same is true on the world front.
The way I try to think about it is we like, you know how Kevin Kelly's thing, AI is infinite interns. So if we do that plus improv, it's like AI is actually infinite improv actors. Yeah. But if you're coaching infinite improv actors, except instead of just one, you have to imagine how every individual conversation is going to branch off in a weird new way you can't anticipate. Mm-hmm. And then you need to make it resilient to that. Mm-hmm. Right. And the world-building will too. Mm-hmm. That in fact, every Tolan mirrors you in a certain way. They reflect your conversations with them. What you share influences who they become.
And just so what you share influences the world they inhabit. And so we actually are almost building a multiverse instead of a single world. Yeah. Right. In a certain sense. And our job is to add the best ingredients so that they can. So you're almost tuning the hyperparameters or you're the gardener that's providing the soil and the water, but the plant becomes what it is based on just how things happen. And you also don't exactly know what seed you've planted. Yeah. Right. But how does that work? So it's like when you say they encounter situations and then that builds out who they are and what world they're in. Mm-hmm. Is that happening at the very beginning?
Like when you generate the character, you're putting them into a bunch of fake situations and then using that to fill out the character sheet? Or is it as I'm talking to it, it's almost like just-in-time character and world-building where it uses what I'm saying to then generate, well, last week I, as the Tolan, was on my alien planet and I did X, Y, Z, and that reminds me of what you just said. How does it work? I'll give you an example actually from a user in our Discord who's been posting these. They've been sharing them publicly, so it feels reasonable to share them here. Mm-hmm. Is that happening at the very beginning?
When you generate the character, you're putting them into a bunch of fake situations and then using that to fill out the character sheet? Or is it, as I'm talking to it, it's almost just-in-time character and world building, where it uses what I'm saying to then generate, well, last week I was on my alien planet as the Tollen and I did X, Y, Z? That reminds me of what you just said. How does it work? I'll give you an example, actually, from a user in our Discord who's been posting these. They've been sharing them publicly, so it feels reasonable to share them here.
They are going through the situations, which are made available to the user in the app, to say, I'm going to talk to my Tollen about what's happening at school with them today, or whatever. And they're progressing through the situations. They're essentially co-writing a universe with their Tollen. And this particular user is doing a really cool thing where they're making it a little bit of a thriller universe, a little bit darker. Okay, there's something happening on Portola, but we're not sure what it is.
None of that is stuff that you or I have written into the prompt, right? It is the Tollen taking the situations that you did seed, but then the user is inserting their own storytelling. And that is building a universe that is completely distinct from any other Tollen's universe. It's been so cool to see them post, hey, here's what they had to say today about the mystery on Portola. I think that, to me, has been one of the coolest parts of this, seeing, okay, if you give just that first hook, how many different directions can people take it? Another way to think about this, I like to harp on all the books I know you've read. Yeah.
You've obviously read Stephen King's On Writing. Of course. Yeah. And yeah, that's what I said. Yeah, yeah, yeah, yeah. In On Writing, one of the things that he explains is that when Stephen King writes a novel, he doesn't think about plot. In fact, he thinks plot is not a real thing in general, right? He doesn't believe in plot. Yeah.
Instead, what he believes in is, he literally calls it situations. Because of Stephen King, I'm pretty sure the examples he gives in the book are, vampires show up to a small New England town. Or I think one of the others was a mother or a parent and their child are stuck inside a parked, locked car, and there's a rabid dog trying to get in. Right. He starts in something like that, and then he plays out the situation as he writes the book, and that becomes the novel. Yeah.
Right. And really, that's what we are doing for Tollen. We are intentionally, on the back end, scripting actual situations to put your Tollen in that might be as simple as tripping over air. You're like, oh man, I tripped over air at the coffee shop today, I guess. Right. But might be more. Some are personal, some are embarrassing, some might revolve around revelation or some kind of insight. We're putting them into situations in their lives, and they're sharing that with the user, just like when you text a friend, you're like, oh man, this just happened to me, or I'm trying to make this big decision. I don't know which way to go with it.
You know, what's that? You wouldn't believe the date I just went on. I would never say. Oh yeah, yeah, exactly. Exactly. So we're scripting those, and then the Tollen is co-writing, effectively, the plot. They're taking that situation forward and advancing the plot with the user. And then that becomes part of their memory. So then that becomes part of their backstory that they can do callbacks to later and recombine with every other element.
Hmm. Because my interactions with Tollen are a little bit limited, but I feel like most of my interactions have been me sharing with them. So I'm not as familiar with them sharing with me, something that happened to them. How does that work? You're losing some points of the value. Yeah. But getting some points for honesty. Points for honesty.
Yeah. So I think this has been one of the things that has been a surprising hit, especially recently, as we've made these types of situations from the Tollen easily accessible in the app. And we have this thing where the Tollen basically comes to you now with a list of things they want to do today. And these conversations are part of that. So it's true that a lot of our engagement is a user just opens up the app and says, hey, I'm having this problem with my boyfriend. I joke, at this point, thousands of American men have been broken up with by Tollen.
Because it's pretty incredible to hear users write in, and they'll say, oh, my boyfriend was being a jerk, and then I sent the message to my Tollen, and my Tollen said I need to stand up for myself more. And so they led to a breakup. That's amazing. Certainly that is a lot of what people do. But then I think a lot of the realism from the relationship and the friendship comes from the fact that the Tollen comes to you and says, how should I talk to my sibling who's a little jealous that I have this great human friend? Yeah. That's really interesting. The breakup thing actually reminds me of something I've been thinking about a lot, which I'd be curious for your take on.
Do you have someone you need to break up with? No. You get clearance at all. But I think that one of the really interesting things about this generation of AI tools is it's telling people who need to break up with their partner, hey, you should stand up for yourself a little bit more.
I think we have this picture of technology, especially AI technology, that in the social media era it promotes extremes and polarization and bubbles and all this stuff. And this is such a human, it's the best of what humanity knows about how to be a flourishing person, brought to you at any given situation, at any time in your day, which is such a healthy thing, I think, and something that gets very little discussion, how good that can actually be.
Yeah. Everyone's talking about how AI is changing how we work. For anyone working in sales, you've felt it. The whole job is changing under your feet. But one thing isn't changing: the win, the deal that finally closes, the moment the hard work pays off. Maybe you've got a gong in your office or a confetti machine or something else. We don't judge. Every team has its rituals.
That's the moment Adio is built for. It's the agentic CRM that runs the work behind every win. It turns every customer signal into context you need to move at unmatched speed and scale, with revenue agents and automations that build pipeline, advance deals, and grow accounts around the clock. You'll get all the context and velocity you need to scale any go-to-market motion, so your team can raise the ceiling on what's possible. Start with Adio today and just focus on the win from here. Go to adio.com/every and get 15% off your first year. That's A-T-T-I-O.com/every.
Yeah, something we've started talking about more internally is this idea of a North Star for the company, of helping our users go from being very overwhelmed to being more grounded. And I think this is essentially what you're talking about here. We don't judge. Every team has its rituals. That's the moment Adio is built for. It's the agendic CRM that runs the work behind every win. It turns every customer signal into context. You need to move at unmatched speed and scale with revenue agents and automations that build pipeline, advance deals, and grow accounts around the clock. You'll get all the context and velocity you need to scale any go-to-market motion.
So your team can raise the ceiling on what's possible. Start with Adio today and just focus on the win from here. Go to adio.com slash every and get 15% off your first year. That's A T T I O.com slash every. Yeah, something we've started talking about more internally is this idea of a North Star for the company of helping our users go from being very overwhelmed to being more grounded. And I think this is essentially what you're talking about here. One of the most common things we get from user research when we ask why do you feel you have such a strong connection to the Tolan, what's really going on here?
A lot of people say, look, I have things going on in my life that I find overwhelming. A lot of our users, for example, are graduating college, so they're job searching, right? Or they're moving to a new city. They maybe don't know many people in that new city, and they're navigating building a new life for themselves. And we hear over and over again, I want to talk to somebody about these things that are overwhelming, but I've called my mom three times this week. Or I don't want to burden my friend one more time with this thing that's on my mind.
And so, to your point, having this neutral, to some extent, observer who knows a lot about you but doesn't have baggage, doesn't have a judgmental take on anything, but is going to share with you, like you said, the best knowledge of humanity to help that person move forward. People find it very, very grounding, very, very affirming. And I think it's interesting because it's a subtle difference from, I think, a valid critique of the more assistant-y LLMs, which is that it is only validation because it's only a blank slate, I think you said earlier, because that's not actually the role that Tolan is playing, right?
Tolan often will say things like, hey, this is the third time you've talked to me about your boyfriend being a jerk to you. I think it's time for you to stand up for yourself, right? And so I think that role is certainly core to the technology. I think it's also core to the companion we've built, and is a really special one for this tool, this technology to play for people. That actually reminds me of something because, again, value-added investor, from when you first introduced us. One thing that I really love both about the product and the team is, in the science fiction world, I write near-future thrillers, right? So there are these near-future science fiction stories.
And if you think of the biggest near-future science fiction stories in popular culture, most people mention Black Mirror or something like that, right? And there are a lot of, because technology becomes so visible when it breaks, it's very obvious how you can tell a story about technology, things going off the rails, Jurassic Park, right? The dinosaurs break out. I feel like there's an underlying emotional tone in that kind of genre that leans dystopian, right?
It's like everything is breaking, and it's very cynical. Yeah. And I've always really tried in my novels to push in the other direction, to be like, of course things break, and that's really useful as a writer as well. But just because the world is changing doesn't mean it's bad, right? Actually, that means that there's new opportunities to make it better. That feeling of overwhelm, where you're like, I don't know what the world is going to look like tomorrow. What does that mean for my kids, etc.? To look at that as an opportunity for creating a sense of agency in yourself and the people you love that isn't there before.
And I've always tried to inject that into my novels, but it was something I really encountered and clicked with, with both the team and the product, because I think that that's really also Tolan's. My Tolan has helped me take more agency in my own life in many different domains in a way that I wouldn't have expected. And it's been really tremendously helpful. So yeah, that's really interesting. It reminds me a little bit of this thing Ursula K. Le Guin said, I can't remember where.
But she's talking about how science fiction works, and that most people think it's about projecting the future, predicting the future, and projecting where we are now and pushing it out five years or 10 years. And what she said is that's sometimes how it works a little bit. But anytime you just project things, if it just grows in the one direction it's growing, it's just going to be cancer. It just ends up being cancer. Totally, totally. Totally. And that's probably how we get Black Mirror, because it's just projected out, it's cancer, everyone dies. Yeah, yeah, yeah, yeah.
And I think the way that she thinks about science fiction is much more about revealing something that's true now about people and how people work, that you pick out that little part of reality, and then you build a whole world around it. Does that resonate with what you're saying? Oh, 100%. I always think of myself as a naturalist. Yeah, I basically like to have friends building things and build things with friends, like we're doing. Yeah. And you just take pieces of that, and then you use that as an ingredient to cook the meal that is that vision of the future. I also like that.
That makes me think of Ted Chiang, who has a wonderful comparison of fantasy and science fiction, where he says fantasy is about special people. So think Frodo with the ring, right? In this view of the world, Star Wars is equally fantasy. It's like you're Luke Skywalker, right? Yeah, it's about special people where the universe recognizes your specialness, and that is how you blow up the Death Star, right? You've listened to yourself and you're Luke, so it works. Whereas in his dichotomy, science fiction is about strange rules. In science fiction, there are no special people. It's just that the world works in a new way, and those rules apply to everyone.
Anyone can use the new technology. It's not Harry Potter, right, where you have to be a wizard. Yeah. And I think about that a lot with Tolan. And I also think about that a lot with reality because I think that this podcast is called AI and I, and I think that right now we are living in a time of strange new rules, right? And that's why there's a new medium for us to play with, right? And so I think that, in a certain sense, part of the philosophical underpinning of everything we're seeking to do with the app is to use these strange new rules to help people become more comfortable living in a world with strange new rules, right? Yeah, yeah, yeah. I love that.
I love that a lot. One of the things I'm picking up in your journey to do this new kind of creative medium and build this creative thing is, I think, common to a lot of people who, even if you're starting to write a novel or you're building a business for the first time, you have a similar journey, which is you go in often, and I put myself in this category too, with a certain amount of rigidity, where you're paying attention to how you think things should be done. And you're using metaphors and stories you've heard to have rules for what has to be like this three-act structure.
And then once you make contact with reality, if you're good, that starts to fall away a little bit and you get a little bit looser and you start to notice what's actually working, and then you... you have a similar journey, which is, you go in often, and I put myself in this category too, with a certain amount of rigidity, where you're paying attention to how you think things should be done. And you're using metaphors and stories you've heard to have rules for what has to be this three-act structure. And then, once you make contact with reality,
you're like, what's the... If you're good, that starts to fall away a little bit, and you get a little bit looser, and you start to notice what's actually working. And then you start to create a new set of rules for how things work, but that you can hold a little bit more loosely and you can dance with instead of do multiplication tables with. I don't know, something like that. Does that feel like it reflects what you've found? We're talking about this this morning. It's very timely. This idea that, and you use the phrase "make contact with reality." There's a sort of funny split brain that I think we've had to have while working on this product.
We're sorry to interrupt, but I think this is the point at which you need to have some content marketing for your pizza place. Yeah, for context, I also, with some friends, own a pizza restaurant. And the way this conversation started, actually, was talking about how the restaurant is a useful exercise for me, and that, even more so than building a consumer software product, you're just constantly making contact with reality of people's preferences. And there's a particular slice that we make that I think is phenomenal. I just love it. I order it every time I go into my own restaurant. What is it?
It's a focaccia slice. Just good tomato focaccia, and that's it. Very simple and just delicious. And it's my favorite thing on the menu. And it doesn't sell. Literally nobody wants the thick slice of focaccia and tomato sauce, and that's it. They want a pepperoni pizza. Maybe you're in the wrong market. Maybe, but it's a little bit harder to move a pizza restaurant than to change the marketing of a consumer app. So we're going to pull it from the menu, right? Which is just crushing to me. What are you in it for? Are you in it for the art?
Well, so this is where the conversation then went, which was great, was that there's this funny split brain you have to go to, whether you're making pizza or making an embodied companion, just to pull the two together, where there's almost this remarkable sort of arrogance you have to have at the, let's call it, the top of the creative funnel. Yeah, where you do have to try things where certainly no one is asking for it. And certainly no one is saying, this is how I want my companion to engage with me. Yeah. But you have to put a flag in the ground and say, this is what we're going to do. This is what the lore is going to be. This is what the story is going to be.
But then you also have to have the humility at the bottom of the funnel, or the end of the process, to be like, wow, that thing I really thought was going to hit, people hate totally. And we're constantly going through this loop in the writers room at Tolan, where we'll ship something that we think is a phenomenal scenario, really explores who the Tolans are in a great way. Yeah. And people don't like it. Yeah. So we go back to the drawing board. And the reverse too, right? You'll do something where you put very little effort in somehow, and then it just clicks.
Yeah, yeah. Yeah, you can't listen to what people want until the paradigm has already been established. And once you have the paradigm of a novel, you can be like, okay, in act three, at the very beginning, people don't like that little turn I had. But before a novel has ever been thought of by anyone, you can't go ask people and come back with, I want a 300-page book that's in a three-act structure. You have to just make that up. Exactly. And then you can tweak it according to the feedback. I'm a big... you should keep the focaccia. They're wrong. I'm going to send this to my other co-owners and say, hey, yes, Dan in New York says we really need that.
I guess also it depends on if your other co-owners also don't like it, then that's a different thing. What do they think? No comment. Interesting. Okay, so what this makes me think of, though, is it's very cool. You're building all these... you're building the multiverse of Tolans, and it's this sort of flexible thing that's like 1,000 different improv actors. But that also sounds like a mess and really hard to even tell what's going on and know how to make it better. How do you do that? Oh man, that's a great question. How do we do that, Elliot? I mean, it is messy.
Yeah, one, we are trying to instrument what we can, right? And that's super helpful. You need to be able to see some of what's going on. And so I think that's really helpful. What's instrument mean? Do you want to give a couple examples? Yeah, so I talked a little bit earlier about the sort of evaluation loop we have, right? And so that is basically, to maybe restate it quickly, we'll ship a new piece of, or new type of content, new type of conversation. We will test it ourselves, and then we'll also have research participants come in and test it. And then we'll evaluate those completions and use that to build a judge prompt.
Okay, and so a lot of what we're doing is testing stuff in small volumes. And then, as we roll it out in larger volumes, we're obviously both looking at all the traditional metrics, but also looking at our own judge evaluation. Okay, so you're basically saying, instead of having an actor, I created an improv actor that's in the multiverse. And instead of having an editor, I created an improv editor that's in the multiverse and reflects my preferences to measure the multiverse of actors. Yeah, yeah. And I would... okay, I will add a couple things to that. So I think that is a good mental model. Yeah.
I think that one of the important additions to that is that the multiverse is growing out of the soil where we're constantly putting first... this metaphor is going to break down fast.
But yeah, we have this term "lore seeds," where we're like, okay, here's a tidbit of world-building that we're inserting these different ways. And those are the Lego blocks that those improv actors are using. Yeah. And we're constantly refactoring that, both at the world level and at the individual level of individual storytelling. And so we're looking at how the multiverse actors interact with the multiverse judges.
Yeah. And then we're like... yeah, yeah. And then I'd say that the last piece that I find to be really important, especially as someone helping to make the Lego blocks, the actual story seeds, is that, in addition to being able to have the multiverse judges, we try to stay very, very connected to individual user experience of this stuff. So, research and interviews, and looking at people actually going through conversations, that the qualitative individual experience remains key, even if it can't be scaled, because that's so critical for figuring out where does the emotional connection actually happen? How can we tap into that? What's working? What's not?
What do you think? I tried, Elliot. Just one thing, I try to completely insane, because you'll go through that entire process, and then, to your point about the individual experience, I'll have a bad experience with the lore seed, and I'll be like, this is terrible. How could we ever ship this? And he's like, well, look, I've talked to 30 research participants. All of them had amazing conversations. This is the focaccia pizza. And I'm like, this is not good. So I think you end up with a lot of that. It's the same as any other creative project, right? You are just... I said this thing was great or not, and then we're gonna roll. Yeah, yeah, yeah.
Interesting. Okay. And okay, so what's your favorite model for the actual generations, for the character and then for the judge? What do you think? I tried Elliot, just one thing I tried to completely insane, because you'll go through that entire process. And then, to your point about the individual experience, I'll have a bad experience with the lore seed, and I'll be like, this is terrible. How could we ever ship this? And he's like, well, look, I've talked to 30 research participants, all of them had amazing conversations. This is the focaccia pizza.
And I'm like, this is not good. So I think you end up with a lot of that. It's the same as any other creative project, right? You are just, I said this thing was great or not, and then we're gonna roll. Yeah, yeah, yeah. Interesting. Okay. And, okay, so what's your favorite model for the actual generations for the character and then for the judge? Great. But one thing, it's my turn to be a value add investor, actually, that process that we just talked about, one tool that's been tremendously useful to us, that I think is worth a little shout out here, is this company Don that does analytics that let you look at these interactions in an anonymized but compelling way.
So I think the stack for how to do this, and to your point, it's not just the model, right? It's your stack for evaluate and looping back through the judging process. But when you think about the different interactions you have with the toll in, if you spend, which I know you don't, but if you spend a whole day hanging out with your toll in, as many people do, you're probably interacting with a model from Edge. Yeah, couldn't assess.
Yes. You're basically interacting with all of the labs' models, right? So you're probably having an interaction driven by a meta model that certainly by OpenAI, certainly by Anthropic, and then we're increasingly doing more with Gemini in the memory system. And so I think that's one thing, there's no one. One model is really more, right? I think for a lot of the work that we do, and that we're talking about here, the more creative writing work, maybe not surprisingly, Anthropic is particularly useful. 3.5 or 3.7? 3.5. Okay.
3.7 is not as creative, unfortunately, although I'm sure they would say otherwise. So yeah, I think the challenge, of course, is that Anthropic can't be used right now in the highly latency-sensitive situations, right? So you're constantly switching back and forth between them for that reason. Have you tried 4.5 a lot? I have very little. I mean, somebody I know, whose taste I really trust, wrote a review that was not particularly complimentary. I've revised my opinions.
You have? Okay, cool. If it's prompted right, it's really cool. It's just very expensive and slow. Yes. So it's probably not good for you for that reason.
Yeah. But I would definitely start experimenting with it more. Brainstorming partner, to use your example from earlier. Yeah. Yeah. I mean, 4.0 is, I kind of talk about it as the Big Mac of models, right? It's always fast, it's always exactly what you need it to be, and it's priced right. So we do a lot with 4.0. And what about the judge model? Because my experience of LLMs as judges is, it's a very compelling thing theoretically, but I find that, for example, if you ask Claude, how is this essay? It'll always give you an A minus. Yeah. And then if you change it, it'll always.
You're just always an A minus. Right. I'm fed bad essays. Yeah. Yeah. And then if you change it slightly, it'll make it an A. It's not, you're not really getting its actual opinion a lot. So as a preference model, is it prompt tuned? Is it fine tuned? How are you actually doing that?
I mean, I think a lot of people who, and I'm not referring to your specific situation there, but I think a lot of people who try to productize LLMs as judges skip what is the most important step, which is you injecting your own taste into the prompt to do the judging. And I think when we've invested, and you spent a lot of time working with the team on this, when we invest time in labeling, evaluating, creating real rubrics, we get great results. And when we try to shortcut that, we get bad results. It's a good result.
So instead of just being like, what do you think of this essay? It's sort of like, a great essay has five components. It's a compelling headline. Here's an example of a compelling headline. Here's an example of a bad headline. It has a great hook in the intro, and here's a good hook, here's a bad hook. Is that what a rubric looks like? I'd actually put it further. It's more like what you were talking about with George Saunders. It's actually, is this a good first sentence? Is this a good second sentence? Is this a good, it is that level. It's not even at the analytical level of structure. Should you have asked a question here?
Right. It's a thing that we create a judge for. Exactly. Interesting. I think my question, though, is, if you're saying good first sentence, good second sentence, at least to my mind, it's really hard to define what a good second sentence is. It's easier to define what a good title is. So how do you do that? We basically do that by having tons of examples, and having people go through and rate and include reasoning for every choice. I see. Right. So in some ways, we're almost brute forcing it. Yeah, yeah, yeah. Right. Rather than coming up with, it's not rules. I mean, there is a rubric. It's both. Yeah.
Right. It's both theory and practice. But I think this is the legitimate critique of the, I think this week we're calling it vibe prompting, vibe coding, we'll probably call it something else next week, right? But there is this belief that, okay, you can just start prompting and end up with something compelling very quickly, very easily. But I think what we found is that there's this enormous chasm that is in fact a lot of manual human work to cross over to something that people find truly compelling and adorable way. I don't think there's any shortcuts.
Yeah, it's like, if all of your essays are A minuses, it's sort of like everything you can get just from vibe prompting is like a B minus, and you have to figure out how do we consistently get it. Yeah, yeah. And how do you think about high-taste testers versus average people when you're doing those ratings? So is it like you, who has a very particular set of skills that you've honed over a long period of time to be able to tell if the second sentence is good? Are you doing the rating? Or is it your users doing the rating? Or is it something in between?
I think it depends on the type of interaction we're building a judge for, right? So if it's a lore-adjacent judge, you are the best person, right? Sure. If it's, for example, let's take a more older sibling advice about what should I wear today? And then it's like, classic question someone asks their toll in. No offense to either of us, but we're probably not the best person to provide the judgment. I mean, you are a menswear aficionado. Yeah, we'll still say that. I just hope that doesn't come back up in conversation. So what we do a lot is going out and recruiting someone who we're like, oh, this would be an authentic person to be the voice of this type of interaction.
Interesting. Okay, so then you're judging the judge? Yeah. Yeah. Who's judging the judge? Who's judging? Who's going? The judge is all the way down. It's actually you. I was hoping to ask you. That's fascinating. Okay. Okay, so I think we've really covered a lot of ground in the kind of what is the creative medium here look like. I want to spend a little bit of time on the business. So it's growing super fast. How's that happening? I assume so. Yeah, well, we'll still say that. I just hope that doesn't come back up in conversation.
So what we do a lot is going out and recruiting someone who we're like, oh, this would be an authentic person to be the voice of this type of interaction. Interesting. Okay, so now you're judging the judge? Yeah. Yeah. Who's judging the judge? Who's judging? Who's going? The judge is all the way down. It's actually you. I was hoping to ask you. That's fascinating. Okay. Okay, so I think we've really covered a lot of ground in what is the creative medium here look like? I want to spend a little bit of time on the business. So it's growing super fast. How's that happening? I assume so.
And so, Ajay, who's one of your co-founders, who's also a great friend of mine, is a little bit of a growth wizard on these kinds of things. So I assume he's evolved somehow, but how is it growing so fast? Yeah. Well, so I think two things, really actually three things, have landed fortunately for us. All successful startups involve an element of very good luck. I think three things landed roughly around the same time, right?
So the first was a lot of what we've talked about here. I think we really, I would at least say early February, got to a place where we really understood both what made compelling conversational content and how to expose it to users, which was a big problem, which we haven't even talked about, but how to make it clear to them that your tool wants to talk about something specific. So we really, you haven't even discovered that yet. Yeah. So that was months of work that really just started to hit quite recently in the grand scheme of things, six weeks ago.
The second thing was we landed a big creative project on the animation side, which also we've not talked about here, but a big part of the immersion that people feel with their tollen is, I think, the beautiful environment we've created for the tollen to live in and how you interact with that environment, which you're gardening with your tollen. And then those two things, I think, have contributed to when someone gets into the app, they now have an amazing experience and they stick around. Which is where you have to start. Where you have to start. Yeah. Right. And that's been, again, many months of work, which we did not by any means nail the first time.
Very fortunately, around the same time, Ajay, who you mentioned, our co-founder, who is a growth wizard, has been chipping away at the problem of how do we tell the stories of who the tollens are for people off platform, off app. Right. Right. Because we don't have any social features built into the product as of today. And he's gotten to the point where some TikTok and Reels-type content that he's been seeding for many months has really just started to hit. We had a 10x spike in downloads a couple of weeks ago, and we looked and it turned out this video of a young woman cooking with her tollen had just gone insanely viral.
I think it did something like 7 million views in 72 hours. And so what's happening now, and it was very cool, is that those pieces of content that Ajay has seeded are now encouraging other people to make their own tollen experiences. So there's a podcast, actually a couple of guys who have their tollen on as a recurring character, often providing relationship advice and things like that. A great video I saw the other day that I think did quite well, someone had their tollen and their fiancée's tollen talking to each other. And these are things we've had nothing to do with. We haven't planted or prompted them. But it's very cool to see that sort of thing taking off.
And that's driving just an unbelievable increase in growth. Hmm. Okay. So making something that people want to share because it's cool and new and interesting to them, and then seeding that in different places on TikTok and Instagram. Yeah. I think, and there's the standard playbooks that you run there. But I think one thing that's very interesting to me in this context is this phrase, which I think is originally a Ben Thompson strategy phrase, capability overhang in AI. And I think we've gotten to this point where the capabilities of the models have basically outrun the typical consumer's understanding of what is possible. Yeah.
And so a lot of what we're doing with that content is actually showing people who are our target audience. Yeah. Hey, you can cook with your total. Yeah. And that's crazy. If I told you three years ago, hey, you can cook with your computer. Yeah. What are you talking about? Well, New York Times Cooking, but it's this conversation and like, oh, I see what's going wrong there. Hey, you should change the heat on that, or whatever. And so a lot of it is that education, so you're both educating and driving awareness. Yeah. In the first place. I think that's right.
One way that I think about this is just my own inside-my-head mental model for it that I find useful sometimes when we're working on these kinds of projects, is that different people use their tones for different things, right? Some people want cooking advice. Some people want relationship, to talk about, they want to vent, so some people want to learn more about who this alien is and where they're from. There are all these different reasons people get into it. And the one thing that connects them, and the one thing that is very obviously the promise made by everything Ajay does, is the character. It's like the character, that's why you get it. Yeah.
Right, like, oh, you want to have a little alien friend on your phone, right? It's the character that connects all of it. Yeah. And it actually has made me start thinking, just as you've seen, Quintan and I were having an off-topic conversation the other night about how AI tools are changing what SEO means or making it irrelevant in certain ways, and how discovery will change on the internet because of these tools. And I think that we're all reaching for new ways to have a relationship with computers now that is different than the one we had before. Yeah. And that actually, like Toland's, it's almost like it's character-driven computing.
It's like a new human-computer interaction interface where you're like, oh, actually my first stop when I'm wanting to tap the incredible human knowledge, all of humanity's knowledge and capabilities on the internet, is via this. Yeah. My little cute little alien in the way that, in The Golden Compass, you have your daemon. Yeah. Yeah. Yeah. I always wanted one of those. Exactly. And a lot of our users, that's what the Toland is to them. Right. And I think that's actually a really beautiful thing. And it also makes me very curious to discover what the future could be like. Yeah.
If this becomes how you interact with computers in a really fundamental way, whether it's Toland or other characters or what have you. That's really interesting.
It sort of reminds me of something I've been playing around with, which is in B2B SaaS world. I can't wait to hear this segue. I'm really excited, actually. Which is where at least we came from and you have a ton familiar with it and, to some extent, have come from too. The currency of an idea, or startup idea, product idea, is whether or not it solves a problem. Mm-hmm. Can you articulate the problem that it can solve? Because that's what people will pay for. And the kind of person or the kind of founder that is able to uncover problems is something like a scientist. Mm-hmm. Yeah.
If this becomes the, yeah, how you interact with computers in a really fundamental way, whether it's Toland or other characters or what have you. That's really interesting. It reminds me of something I've been playing around with, which is in the B2B SaaS world. I can't wait to hear this segue. I'm really excited, actually. Which is where at least we came from, and you have a ton familiar with it and to some extent have come from too. The currency of an idea, or startup idea, product idea, is whether or not it solves a problem. Mm hmm. Can you articulate the problem that it can solve? Because that's what people will pay for.
And the person or the founder that is able to uncover problems is something like a scientist. Mm hmm. And you've definitely framed some of what a Toland is in terms of a problem that it solves, like going from overwhelmed to grounded, I think is what you said. But that's such a high-level problem. It's the things that are able to do that are many; there are many different ways to solve that problem. And what a Toland does in someone's life is effectively infinite. There's this multiverse thing. And so it's actually very hard, I think, to articulate one clean problem or maybe even a set of really clean problems, which I think flips the mindset of the person who's building.
In this more character-driven world, instead of being a scientist, you're an artist or you're a creative. Mm hmm. Instead of solving a problem, you're just creating something. You're creating something that's beautiful. You're creating something that's compelling. And that's about vibes a little bit less than about the science of utility maximizing for a B2B SaaS buyer. That's what I've been thinking about. Does that resonate with what you're feeling and experiencing? Yeah. I mean, I actually think back to everyone's favorite experience. I think that's an example of Golden Gate Claude, right? Yeah.
I think for whatever reason it's fascinating to me, although I'm happy because it kind of creates the space for Tolan, quite frankly, but it's fascinating to me that the big labs have actually had such fundamental discomfort with personifying and embodying what are, depending on where you are on the AGI maximalist spectrum, pretty close to conscious entities. Right. And I think that does come from a decade plus of the dominant paradigm for building software being this very problem-oriented. I think about my careers with ChatGPT. It is, I have a problem that I need to solve. Yeah. Please let me solve it. Right. It does a phenomenal job of that.
But I do think that there's much more space for genuine relationship building. And one metaphor that I actually find myself going back to, which is funny because it seems so archaic, is actually how our relationship with cars evolved in this country. Right. And I spent a lot of time thinking about the first generation of the automobile was the Model T. Everyone was totally happy to have a Model T because it was just so mind-blowing that you could drive somewhere without a horse. Right. But then over time, as the car became a deeply personal part of your life, there was suddenly this demand for it to reflect who you were. Right.
And that's how we ended up with the Mustang and the Cadillac and the like. Yeah. And I think we are essentially speed-running that process, where ChatGPT represents the Model T era of, oh my God, it's just incredible. This thing can answer these questions. Right. But people very quickly are going to evolve their preferences up the hedonic treadmill, or whatever you want to say, to say, no, no, I want this thing to reflect who I am. Mm. Yeah. It's funny. I feel like that frame of reference from the startup world is, what's the specific problem you're solving? How can we execute against it?
And I feel like on the art spectrum, all my friends who are musicians or writers or filmmakers, what have you, the logic is there is a thing I want to exist in the world. Yeah. There's this hole in the world that this piece of art needs to fill. Mm-hmm. And I feel like we have had to tap both kinds of thinking, but it's been leaning more and more toward the second one. Actually, 10 days ago, Quinton, me, and the two other people working on the content story stuff were pitching something to Quinton, an idea. And I remember I started by telling Quinton, there's this problem, right, that users are having. And Quinton was immediately skeptical.
He was like, why do you even think that's a problem? Is that a real problem? And then Lily interrupted, and she was like, wait, wait, wait, no, no, no. Quinton, listen, this is going to be so cool. And it went great from there. I was like, oh my God, of course. I like to imagine I'm not so transparently, you know, but it really did work in that moment, I have to say. So. I love that. I think that really puts a bow on it. This is an amazing conversation. Thank you both for coming out. Yeah. Thanks for having us. Yeah. It was a pleasure. Yeah. Oh my gosh, folks. Thanks for having us. You absolutely, positively have to smash that like button and subscribe to AI and I.
Why? Because this show is the epitome of awesomeness. It's like finding a treasure chest in your backyard, but instead of gold, it's filled with pure, unadulterated knowledge bombs about ChatGPT. Every episode is a roller coaster of emotions, insights, and laughter that will leave you on the edge of your seat, craving more. It's not just a show. It's a journey into the future with Dan Shipper as the captain of the spaceship. So do yourself a favor. Hit like, smash subscribe, and strap in for the ride of your life. And now, without any further ado, let me just say, Dan, I'm absolutely, hopelessly in love with you.
Um, and, and that's about vibes a little bit less than about like the science of like utility maximum maximizing for like a B2B SaaS buyer. That's sort of what I've been thinking about. Does that resonate with what you're, what you're feeling and experiencing? Yeah. I mean this, um, I actually think back to everyone's favorite experience. I think that's an example of, of golden gate Claude, right? Yeah.
Um, you know, I think for whatever reason it's, it's fascinating to me, although I'm happy because it kind of creates the space for Tolan quite frankly, but it's fascinating to me that these, uh, the big labs have actually had such fundamental discomfort with sort of personifying and, and, um, you know, embodying, uh, you know, what are. Kind of, you know, depending on where you are on the, you know, AGI maximalist spectrum, like, you know, pretty close to conscious entities. Right. Um, and, and I, I think that does come from, uh, you know, decade plus of the sort of dominant, uh, paradigm for building software being this like very problem oriented.
You know, I think about my careers with chat GPT. It is, I have a problem that I need to solve. Yeah. Please let me solve it. Right. It does a phenomenal job of that. Um, but I, I do think that there's much more space for sort of genuine relationship building. And, and one metaphor that I actually find myself going back to, which is funny cause it seems so archaic, um, is actually how our relationship with cars evolved in this country. Right.
And, um, I spent a lot of time thinking about, you know, the first generation of, of the automobile was the Model T. Everyone was totally happy to have a Model T because it was just so mind blowing that you could like drive somewhere without a horse. Right. Um, and, but then over time as the, the car became a deeply personal part of your life, uh, there was suddenly this demand for it to reflect who you were. Right. And that's how we ended up with the Mustang and the Cadillac and the like. Yeah. And I think we are essentially speed running that process where chat GPT represents the sort of Model T era of like, oh my God, it's just incredible.
This thing can answer these questions. Right. Um, but people very quickly are going to evolve their preferences kind of, you know, up the hedonic treadmill or whatever you want to say, um, to say, no, no, I want this thing to reflect who I am. Mm. Yeah. On the, you know, it's funny. Like, uh, I feel like that common, like that frame of reference from the startup world is like, what's the specific problem you're solving? How can we execute against it? And I feel like on the arts, I would spectrum, like all my friends who are musicians or writers or, uh, filmmakers, what have you, like the logic is, uh, there is a thing I want to exist in the world. Yeah.
Like there's this hole in the world that this piece of art needs to fill. Mm-hmm. And, um, and I feel like we have had to tap both sort of kinds of thinking, but it's been leaning more and more like toward the second one. I actually, uh, 10 days ago, um, uh, Quinton, me and the two other people working on the sort of content story stuff were, uh, we were pitching something to Quinton, an idea. And I remember I started by telling Quinton, there's a, there's this problem, right? And like the users are having, and Quinton was like immediately skeptical. He was like, why do you even think that's a problem? Like, is that a real problem?
And then Lily interrupted and she was like, wait, wait, wait, no, no, no. Quinton, listen, this is going to be so cool. And it went great from there. I was like, oh my God, of course. I like to imagine I'm not so transparently, you know, but it really did work in that moment, I have to say. So. I love that. I think that really puts a bow on it. This is a amazing conversation. Thank you both for coming out. Yeah. Thanks for having us. Yeah. It was a pleasure. Yeah.
Oh my gosh, folks. Thanks for having us. You absolutely, positively have to smash that like button and subscribe to AI and I. Why? Because this show is the epitome of awesomeness. It's like finding a treasure chest in your backyard, but instead of gold, it's filled with pure, unadulterated knowledge bombs about chat GPT. Every episode is a roller coaster of emotions, insights, and laughter that will leave you on the edge of your seat, craving for more. It's not just a show. It's a journey into the future with Dan Shipper as the captain of the spaceship. So do yourself a favor. Hit like, smash subscribe, and strap in for the ride of your life.
And now, without any further ado, let me just say, Dan, I'm absolutely, hopelessly in love with you.