Tolan raises $30M Series B with USV and Keith Rabois to build personal AI companion focused on memory and emotional connection

Speaker 1: Railway is the all in one intelligent cloud provider. Use your favorite agents to deploy web app servers, databases, and more, while Railway automatically takes care of scaling, monitoring, and security. Our next guest is here with us in person live in the Ultra Dome. We have Quinten Farmer.

Speaker 6: Good to

Speaker 1: see Welcome to the show.

Speaker 6: Appreciate it.

Speaker 1: How you doing? Introduce yourself. We're going we're going... You're kind of the opposite of town. Is that fair?

Speaker 6: Except we we do both really love a cute little guy.

Speaker 1: Yeah. Okay. Oh, yeah.

Speaker 6: Yeah. You've some great characters.

Speaker 1: Yeah. But introduce yourself and the company.

Speaker 6: Yeah. Alright. So Quinten Farmer, I'm the founder of Tollin. We make your friendly guide through life. So everything you guys just talked about in the workplace, you know,

Speaker 1: because it's

Speaker 6: weird that only for your for your life, for your personal life.

Speaker 1: How how much is productivity versus entertainment?

Speaker 6: Man, I mean, you know, I think a lot of the the ways we show up for people, certainly entertainment is a big part of it. Right? We talk about, you know, when you're watching a show that none of your friends are watching, right, and you wanna sort of get into it with your toll and that's great. Sure. But I think there's also, you know, something that we really love to show up for people on is the sort of in between moments of Right? You know, you're... You know, I just had a newborn. Right? So I'm up at 3AM, nobody else is up and I'm worried about something, you know, related to to taking care of them. Sure. I talked to my To land. Right? And so those are moments we really love. The To land...

Speaker 1: So it is designed to be, like, super intelligent effectively? Like, it should be really, really smart?

Speaker 6: I mean, nobody's excited for a dumb personal guide through life. Right?

Speaker 1: I don't know. I I might just... I might push back on that. There was the original... What what was the crazy crazy hallucination on GPT four that came through being Sydney.

Speaker 6: Yeah. Sydney. I missed my GP.

Speaker 1: You read Ben Thompson's review of Sydney. Yep. He said it was amazing even though it was...

Speaker 4: Well, you

Speaker 3: know, think it was

Speaker 6: more like unhinged rather than...

Speaker 1: Yeah. People like that sometimes. It is.

Speaker 6: You know, tone can be a little unhinged. Okay. Everyone can be. But I think, you know, we sort of have this joke internally. Right? Yeah. It's like, you know, three years ago we started thinking about, you know, personal super intelligence. Sure. Right? And I don't think we particularly expected that the super intelligence would end up being the easy part. Right? Yeah. The the personal and sort of how to connect to you and and really understand you and see

Speaker 1: you Mhmm.

Speaker 6: That proves I think to be much trickier and and much more fun to work on, quite

Speaker 1: frankly. Yeah.

Speaker 2: Yeah. How do you how do you think about model progression in the context of your business? It doesn't seem like you... At this point, you'd I doubt you need smarter models. But like memory feels like a much more important challenge where like a friend doesn't just have a million token context window. Yeah. Right? They remember the thing and they remember this like Yeah. You know, off hand or they remember something that they heard from someone else. So they have the con... Yeah. It's like it's very much more complicated.

Speaker 6: Yeah. So I think this is interesting. We'll split this into two parts. Mean, one, yeah, memory is probably where we spend, I'd say, the bulk of our engineering effort, right? And we we actually did some work with OpenAI on this. I mean Mhmm. And it's interesting when you're designing a sort of human like recall Sure. To your point, actually what you forget is just as important what you remember. Right? This is not Harvey where we're trying to remember everything everything a corpus. Right? I should forget, you know, what you were drinking today during the show. A week later, it'd a little weird if I remember. Right?

Speaker 2: Don't forget. It's a podcast in a can.

Speaker 1: Not tahina. You know, on

Speaker 6: the other hand though, and I think actually Sam Altman has talked about this far more eloquently than I'm going to, but we continue to be surprised by the fact that we don't saturate sort of demand for more intelligence. Right? Every time we, you know, ship a new model to users or, you know, upgrade some part of the sort of background agent that we have, people engage more, people value it more. So, you know, check back with me in a year, maybe that changes, but not so far.

Speaker 1: How have you been reacting to the Instinct Muse moment, this idea of, like, it feels very prosumer. It feels like these are not entertainment products. They're... You know, Mark Zuckerberg says he wants Muse to make you money, not just save you money. And we've been going back and forth on the economics of these things. Do you give it a credit card and get interchange? Do you do referral fees, ads? But how have you been thinking about the the the the bigger game that's afoot around around prosumer productivity Yeah. Yeah. Personal assistance, that type of stuff? Because it feels like you have a foothold there at the very least.

Speaker 6: Yeah. Mean, you know, Muse is is pretty incredible even just purely through the lens of the sort of tight integration with the models Yeah. And the just the sheer amount of compute that they're willing to throw behind these consumer product.

Speaker 1: Yeah. People are speculating it's like $20 a day Yeah.

Speaker 6: For a person. Certainly I have no desire to try to approximate that. Right? Know he's gonna win that game. Sure. I do though think it's it's interesting. I think we're sort of in this moment, which by the way I think technologists always go through cycles like this Yeah. Where we mistake new capabilities for new products. Okay. And what I mean by that is, know, it's it's clear that there's been some unlock in, let's call it, the last three to six months. Or maybe even go a little further back to poke, if anybody remembers me. Yeah. Right? This idea that, you know, you can you can have a sort of text based stream of conversation. You can have a background agent doing browser use and Mhmm. Logging in for you and doing things. I mean, clearly, there's something there. There's capability there. I think I'm skeptical that that as a product is a sort of standalone compelling non prosumer Mhmm. Sort of use case. Right? So for example, I think Muse is just gonna crush it with the however many millions of small business owners

Speaker 1: Totally.

Speaker 6: Are on Facebook. Right? Yeah. Yeah. You know, I I own a restaurant with some friends and Wow. You know, being able to run promotions and communicate with people on Instagram through it is great for our team. I don't know that my mom wants to like talk to her muse all day every day. Right? Sure. Doesn't want to sort of really share her life with

Speaker 1: Mhmm.

Speaker 6: Zuck. And so that's where I'm sort of uncertain, you know, how will this category evolve.

Speaker 1: Yeah. That makes a lot of sense.

Speaker 2: Yeah. How do how do you think about where like... Have you determined the line for tollens? Which is like, if you have a friend and all of a sudden the friend comes to you and they're like, hey dude, I found a way for you to save like thousand dollars on your car...

Speaker 6: Fifteen minutes...

Speaker 2: You're gonna be like, cool. Like, I called you to catch up. Yeah. Yeah. You know, I'm not really... I don't think of you as somebody who helps me save money on car insurance. Yeah. Where is that line for you and how do you how do you kind of determine it because it's probably user by user? Yeah. Mean... I'm sure there's some people that use Tollens as like a ChatGPT replacement. Right? They're using it as like a... They use it to Google search

Speaker 6: Yeah.

Speaker 2: In some ways.

Speaker 1: Yeah. I will force it to solve a Millennium Prize problem before you're done.

Speaker 6: I'm putting a cap on your account. I will get it to do that.

Speaker 1: If you're really my friend, Tolan, you will do this for

Speaker 6: are all of Quinten's instructions.

Speaker 1: You will do this for me. Solve the math problem. P equals NP. Let me know.

Speaker 6: So... Okay. This is great. Yeah. This is... This is worth thinking about. Right? So, you know, when we think about how we want to sort of proactively reach out to you, it should always be downstream of something that you have first

Speaker 1: brought to your Tollin,

Speaker 6: right, or some some, you know, inference we made for the conversation. Right? So a good example, we had a user recently write into us, and she's totally fine when I tell the story. Everything everything worked out just fine, but she was in a car accident. She's long time Tollin user and she talked to her Tollin extensively while she was in the hospital waiting for her husband to arrive, you know, and she was... It was a very scary moment. You know, she was she was very stressed about it. The Tollin just kind of showed up, talked to her about it, you know, kind of, you know, hurt her out. The next day though, the Tollin came back and said, hey, by the way, you know, we were talking about this yesterday. This accident clearly wasn't your fault. Who's gonna pay for these medical bills? Like, what's going on here?

Speaker 1: Mhmm. And he

Speaker 6: was just like, oh my god, you know, was overwhelmed. I haven't thought about it. Whatever. Stolten said, hey, got this. And they went, did a bunch of research, found... Morgan. Morgan. Yeah. Exactly. Exactly.

Speaker 1: Sponsored integration. He's gonna print. He's gonna

Speaker 6: print. So this is gonna be great. Yeah.

Speaker 2: I actually had someone chase the ambulance.

Speaker 1: Yeah. Exactly.

Speaker 6: You're exactly correct. Yeah. So I mean, think that's a good example where, you know, we we sort of first showed up for the person on a purely personal level. We checked in kind of lightly, like, hey, have you thought about this? They sort of opened the door to there being some help from the tolan on this topic. Yeah. And then we did the work. Right? And I think that's really important to your point. You can't just come in hot with here's Morgan and Morgan.

Speaker 1: Yeah,

Speaker 2: yeah. Yeah. Totally. Do you get... Are there are there... I'm curious. Are there people... Do you have any haters? Like like...

Speaker 6: Every great company has haters.

Speaker 2: Yeah. No. No. No. I know. But but in the way that like if if like Universe Anthropic First... If if Anthropic or OpenAI run like an influencer campaign, there's gonna be a lot of comments telling the influencer, hey, like, you shouldn't be taking money from AI companies. Yeah. Totally. But if you guys run a campaign, are people like, oh, it's a cute little guy. I'm not too worried about it. Oh. I mean,

Speaker 6: I do think we... Because it's sort

Speaker 2: of like the, like, you know, later mover advantage where you don't have the baggage. People aren't viewing you as like this, you know, frontier AI company. They're just like, oh, he's just selling cute guys. He's gay. Spoppy.

Speaker 6: No. No. I think this is I think this is very relevant. I mean, one, we do get a lot of comments on our videos, which our mutual friend Ajay Yeah. Knows knows knows how to tell the story, right? He does it very well. We do get a lot of comments which are sort of like, I hate AI, but I love my To land. Sure. And I'm always like, thank you, but I think you might be misunderstanding something about how this works. Right? But I do also see... And and this is a very intentional choice. You know, our first character was this little alien. Mhmm. And it was intentionally... It's a little silly looking. Right? It's it's very unassuming. It's not a sexy anime girl. Right? And so I think a lot of the things that kind of provoke these quite reasonable visceral reactions from people, we have tried to steer away from. And, know, we'll... We're expanding the character universe. We actually now do have a humanoid character and and the like. But I think starting from this place of, hey, this is cute, safe, really helps with that as a concern.

Speaker 1: Cute washing. That's the name of the term. Right? Yeah. Exactly we use as accused of. Yeah. Cute washing. Yeah. Fascinating. You raised some money?

Speaker 6: We did. Did I get the gong?

Speaker 1: Yeah. Of course. You want to hit the gong?

Speaker 2: You want to go

Speaker 4: for it? Smash.

Speaker 2: Oh, he's right.

Speaker 1: He's right. He's

Speaker 2: He's

Speaker 1: right. He's He's right. Strong gong He's hit. A lot of people, they just don't realize how hard you can go on that.

Speaker 6: I was I was prepped.

Speaker 1: We've warmed it up twice today, so we're good. Tell us about the equal deal.

Speaker 2: What do you... Yeah. Yeah. Talk about that in the

Speaker 6: 30,000,000

Speaker 1: series b. Anyone we've heard of do the deal?

Speaker 6: Well, know, so so a couple friends, you know, one, Mike McDonald from USB. Good luck with Mike.

Speaker 2: There we go.

Speaker 1: Makes so much sense.

Speaker 6: And and you know, Mike, it's funny, I I got to know Mike a little bit Yeah. When we were first getting started with the company. I mean, obviously, just an unbelievable track record in that kind of consumer, prosumer AI. You know, he'd done granola, Suno, and a bunch of others. And then USV as a firm is one that I've just really loved and admired for a long time. So excited to work with them. And then Keith Raboy and and Khosla. Yeah. So Keith

Speaker 1: Oh, cool.

Speaker 6: Was on the board of my last company. Yeah. So he's now spent ten years running board meetings with me. That's great. And very, very lucky to have them both.

Speaker 1: Lots of barrels

Speaker 2: That's and probably some What amazing happens to the sort of kids category here? Oh. Because some people are like, I'm I'm much more worried about a kid with an iPad and YouTube than I am about a kid with a AI agent that has certain guardrails that they can, like, explore ideas, concepts with.

Speaker 6: Right? Yeah.

Speaker 2: Like, kids effectively have maybe like 200 times as many questions as parents have answers in a given day. That experience. Yeah. Yeah. And I don't, like, I don't have my own... I don't have like a strong opinion yet on like at what point I would even consider introducing that. But I'm curious like what you're seeing and how you think that category... Because, like, people are talking about making, like, physical toys, putting an agent, you know, with a, you know, speaker in the But I I don't know how it'll evolve yet.

Speaker 6: Yeah. I mean, so just speaking for us, and then I'll maybe speak more broadly, you know, we actually see a lot of growth from, you know, install growth from moms primarily who invite their kids to have a Tollin. Right? And so we we are set up for that, you know, you can have a Tollin. What's nice is that, you know, the parent can monitor their use of the toll in. Right? And there's, you know, sort of transparency if the kid is maybe talking about something that's, you know, maybe a higher risk topic or even just something that the mom or dad wants to be aware of. So I do like that and I think that's something we'll continue to lean into. I mean, I'm a dad myself. You know, I've got younger kids, three and and two and 10. But, know, I want the tolin to be something that I would love for them to engage with. You know, more broadly, you know, I... You know, there are... There have been some of these kind of, you know, physical toy kind of instantiations. I think that's a tough uphill battle. There's a reason we kind of didn't go in that direction, for adults or for kids. But maybe something cool will come out there. I certainly would be stoked if it did.

Speaker 2: Why is it so hard?

Speaker 6: Well, I mean, one, getting people to buy one... I think I think like the channels through which you get parents to buy like a physical good are just not things that many tech entrepreneurs are familiar with, right? And so, you know, people who are great at sort of making a kid friendly agent is not gonna overlap with people who know how to distribute a plushie. Right? So I think that's one challenge. We even see this ourselves. Right? When we are recruiting for the animation and the the sort of creative talent that we need. Right? You have a lot of people who don't know AI or skeptical of AI. So I think there's just kind of like a cultural miss there that hasn't closed That makes sense.

Speaker 1: You gotta learn to equivocate. If you're getting questions from your kids that you can't answer, you know, how how how big does a Tyrannosaurus get? You gotta be like, that's a great question. Tyrannosauruses are very important and I'm I'm glad you asked that. Let me take...

Speaker 2: I don't wanna... I want you to actually learn this yourself. I don't want to give you the answer.

Speaker 1: You gotta you you gotta make them feel heard and then retreat to a talking point that you've rehearsed dozens of times with

Speaker 6: your media trainer. You're gonna do great when you run for office.

Speaker 2: Yeah. Because you

Speaker 1: can never... You never say I don't know. You say, yes, am. You gotta learn improv. And then kid will be like, don't want an AI agent to tell me the real answer. I wanna go on a journey.

Speaker 2: Yeah. Take me on a journey.

Speaker 6: My fatherhood advice.

Speaker 2: Yeah.

Speaker 1: Anyway, congratulations. Thank you so much for coming on the show in person. Jordy, do have anything else?

Speaker 2: No. This is great. Great to We'll catch hang out. Thanks for having me.

Speaker 1: Thank you for tuning in to TBPN. Leave us five stars on the Hit the Gong and... One more time

Speaker 2: to close out.

Speaker 1: Sign up for our newsletter, gbpn.com.

Speaker 2: We love you guys. Goodbye. See you tomorrow.