Andrew Laffoon's very first photobook was for two college friends moving back to Singapore. He invited everyone to send photos and write something meaningful. Almost nobody could. They ended up printing the book blank and passing it around like a yearbook so people could write in it by hand.

After early product-market fit, Mixbook grew fast — $80K, then $1M, $4M, $15M, $25M. Then it stalled for five years. Laffoon now says the worst advice he ever took was treating his exec team as his mentors. Someone who reports to you has too much riding on staying in your good graces to tell you the hard truth.

Laffoon redefined humility not as low self-regard, but as having high regard for yourself and for others at once. That intersection creates the self-awareness to learn from both yourself and others — and a far faster growth rate. He then rebuilt Mixbook's operating model around Jim Collins' flywheel concept from Good to Great.

Breakthrough AI Operators — Episode 207 | Guest: Andrew Laffoon, Co-Founder & CEO of Mixbook

Andrew Laffoon (00:00): Storytelling is not niche. Taking photos is not niche. So why do photo books need to be niche? It's just that they're too hard and people haven't identified with a solution.

Roland Siebelink — Host (00:43): Welcome back to Breakthrough AI Operators. This week, I'm so excited to have a guest who barely made it onto the show after huge delays with his travel, and he just prioritized his schedule to be on our podcast today. I'm super proud. Now let me see if we can guess who this guest actually is. Today's guest started coding at age 10. He got to Inc's 30 Under 30, most promising CEOs under 35, and then he co-founded his company already 20 years ago. That's really rare in the Breakthrough AI Operator space. And that was back when photobook meant fighting with a template, and he's still the CEO — the one who stayed, same chair. Now here's the unusual part. Most AI transformation stories come from two-year-old startups with no legacy and lots of compute. Today's guest is doing the opposite. He's taking a 20-year-old consumer business, millions of customers, four continents, profitable, and rebuilding it end to end on generative AI. He launched a feature only last week that lets you describe a memory in plain English and gets a finished photobook back. He told me he cried reading some of the things that it wrote for his personal photobook. And with that, everyone meet Andrew Laffoon. He's the CEO and co-founder of Mixbook. Welcome to Breakthrough AI Operators, Andrew.

Andrew Laffoon (02:15): Thanks, Roland. Glad to be here.

Roland Siebelink — Host (02:17): Andrew, let's dive right in. You said something in our prep call that stopped me. "Writing is an act of vulnerability, and humans are bad at it." For 20 years, the photobook industry optimized for templates and price and design and nobody addressed the real blocker. How did you see that real blocker when the whole industry missed it for so long?

Andrew Laffoon (02:40): Yeah, interestingly, my very first photobook was a book for two of my best friends from college. They were graduating, they were moving back to Singapore, and I had this idea. We always used to get together before friends would leave the area and we would encourage them — also make fun of them, all the things, all the friends get together, share stories, laugh, cry. I had this idea: what if we could turn that experience into a book? How cool would that be? I invited all my friends to send me photos. I got one CD of photos — remember, 20 years ago — and I invited my friends to share some of their encouragement. I didn't really get that much. There was no way to do it online. I tried putting it in the book; you couldn't really put it in, it was very difficult to do. What we actually did was we printed the book, brought it to that experience, and had everyone write a note, like a yearbook. The oldest way of actually adding your text in a book. Fast forward 20 years, it's still a struggle, even as far as the tools have progressed, to write meaningful sentiment. Even saying "meaningful sentiment" is so hard for humans. But when you can do it — what does it do? It builds connection, it builds relationship, it adds perspective. People see blind spots they didn't even know they had and feel loved and they feel cared for. They feel part of something. That's been a big part of our passion — it's storytelling and connection together that actually makes the difference.

Roland Siebelink — Host (03:37): Tell me a little bit about how the company's grown over the years, how many people are in your team today, what growth have you seen lately?

Andrew Laffoon (04:13): We started the company, as mentioned, 20 years ago — actually this week, funny enough, so this gets to be our official 20th anniversary podcast. We got to product-market fit almost effectively in scrapbooking. Our customers were people who actually want to write the text, who want to design the layout, who want to pour in all the detail. And in fact, they come home after work and they tune out, they get on their computer, and they Mixbook for fun. We built this hobbyist audience that was a great foundation for a business, but it was hard to cross the chasm. There are these unique people who are so caring and so particular in what they do, and they're willing to take the time to develop all the skills to make it the way they want. But most people are not like that. We've basically been spending the last 10 years on: how do we cross the chasm with this need that everyone feels, but make it accessible for everyone? What we've been on about is: can we actually bring back the ability to create something deeply meaningful where your story is actually there? And I think as we've gotten closer to that, the business has grown, the audience has grown, our ability to capture people's interest has grown. But photobooks basically are still viewed as a very niche product. I don't think they should be. Because storytelling is not niche. Taking photos is not niche. So why do photo books need to be niche? It's just that they're too hard and people haven't identified with a solution.

Roland Siebelink — Host (05:48): Of course, this is 20 years later, but you just launched a new AI version of your product. Tell me more about that launch and how it's been showing some initial traction if you can.

Andrew Laffoon (06:01): We're eight days in — we launched Story Mode last week. We talk a lot internally about this being the beginning of the next 20 years. We finally have the tech to cross the chasm. Thanks to OpenAI, thanks to Claude, thanks to Google — these guys have actually pioneered a tech that makes it possible to organize your photos in a way that reflects your intent and to create a book that, honestly, 98% of people could just not make on their own. One of the first ones I did was a book for my wife for Mother's Day. This was actually an earlier version of the tech. You still have to pick the photos right now — we don't yet pick the photos for you, that's coming soon. But you pick the photos and then you give it a prompt: make a book for Mother's Day, walk through different parts of our journey together, walk through things I love about her. Obviously prompting is still a thing — you need to prompt this thing. This is ChatGPT for your photobooks. And the more you prompt, the better it's gonna be. And it laid out the whole book and it weaved this narrative on every page. And it did it poetically in a way that for me to write it like that, yeah, I could have probably gone back to ChatGPT for every single page, which would have been a pain. Most new users will give a very simple prompt at first — they'll say "make a book of my last trip." I'm sorry, that's just not that useful of a prompt. So it will then say, okay, tell us more about this trip. And you say, we took a trip to Greece. Who was with you on this trip? Because once you tell it more, and the more you tell it, the better. It'll get to the point where it thinks it's good enough — it won't take you to every detail because then people will start to drop off.

Roland Siebelink — Host (07:43): Right, exactly. Okay, that's very good. And what was the hardest part in developing that tech? Was it trying to tap into those emotions? Understanding what information you need and don't need? Finding that point where you can cut off because it's good enough, as you just said. What was the hardest part in your experience?

Andrew Laffoon (08:10): Yeah. I think with all products — and AI products maybe in particular — that good-to-great part is really, really difficult to navigate. Because it gets to the point where it's pretty good and you could very easily decide, let's just move on to the next thing on a roadmap because we have these crazy exciting things in the future. But we recognized there was a spark in this product, and if we could just get it to the wow moment. My Greece book — the book for my wife — as I was reading through the narrative, Roland, it actually made me emotional. I started tearing up. My wife was just working on our book from our Greece trip, and it was a big milestone for us. There was a lot of personal stuff we'd gone through that this was celebrating us moving past. And the way it narrated that was unbelievable. It's just like, how did it know we were thinking that? And even more than that, it actually recognized feelings and made connections that we ourselves hadn't made. That's the whole sentience of LLMs — you can feel it there. There's this emotional sentience that helps you recognize your own experience differently. And that's crazy to me. Honestly, I didn't know that was going to be possible.

Roland Siebelink — Host (09:28): With that experience, of course, I cannot help but ask the question — how do you now think about this existential question these days? Are AI models conscious? Are they sentient?

Andrew Laffoon (09:43): This could be a podcast in and of itself. I do not think they're conscious. I do not think they're sentient. We could spend a lot of time defining terms. The best person I've ever heard explain this is Andrej Karpathy — he has a way of making it accessible and understandable. Walking through his first deep dive into how LLMs work, he's like: it's basically fancy autocomplete. I think it's beyond that now. When you have chain of reasoning, it's beyond fancy autocomplete. But it is less than what a human can do in the sense that it lacks its own will. And its self-awareness is, in my view, a mirage. The self-awareness persists as long as the context window. But when you're within the context window, it's pretty compelling.

Roland Siebelink — Host (10:12): I wanted to go back a little bit, Andrew, also to what you said about this being a new chapter in the life of Mixbook. You've been at it for a long time, and many of our listeners are newer entrepreneurs, people who still have a big journey ahead of them, hopefully. How do you think of the chapters in a startup's life? What is a new chapter? When is it time to start one? When is it time to finish one? And how has this metaphor helped you and your team navigate all the challenges you've been facing?

Andrew Laffoon (11:03): Yeah, that is such a right question. I think the number one thing for a startup, as a founder, is you. If you stop growing, the company will stop growing. And it may take a little while, but it will catch up. The company at some point will only grow as fast as you can grow. If you're not self-aware, you're gonna be missing major things that are going on in the company and you're gonna get stuck. Once you get to product-market fit, the problems are just so different. And they hit you so fast. We were growing — we did $80K, then a million, then four million, then 15 million, then 25 million — boom, boom, boom, boom, boom. It happened really quickly. And of course, that's nothing compared to the way companies grow today with AI. But back in the 2000s, it felt fast to us. How do you keep up? And I made a lot of mistakes. I didn't bring in mentors at the right time. And on the flip side, my board was like, you need to hire execs who can tell you what to do. And I took that seriously. That was such bad advice. Your exec team is not your mentors. That's not right. You need someone objective, third party, who can actually help you for you. You need someone speaking into your life in that way to help you grow. And then you need an exec team who you can lead and who you can trust — but they're not just gonna go solve it all. There's so many other things that come with it that are not intuitive, especially to an engineer. I just didn't understand even the basics. I didn't really play team sports. I didn't know how to get people to work together — just a skill I didn't have. And I can be persuasive and I can get people to go in a direction, but if they don't like each other, it just doesn't get that far. After we got product-market fit, we grew really fast. And then there was another chapter where we hit a wall. And I knew we hit a wall. I just didn't know how to get around it. And the things I tried didn't really work. It took us basically five years to really get back to where the growth was actually pretty healthy again. You get stuck at a certain point. And it's primarily organizational dynamics — and your own readiness to grow as a leader and admit where you are. And actually say, okay, I do need help. For us, we got stuck. And the way we got unstuck was two things. One, me actually being real about: I'm burned out and I'm done. And I need to actually make a decision. Either I'm gonna go take a break and then go big, or let's just sell and be done with it. You can't every year be trying to remake that decision. Every year, are we selling? Are we growing? What are we doing? You can't live in that tension for so long. Then, as soon as I made that decision — I went on a retreat and I made this decision — I wanna do this because the mission matters to me. I actually care about it.

Roland Siebelink — Host (14:07): Reconnecting with your mission, with your purpose, in a way?

Andrew Laffoon (14:09): Yes. Going back to that original vision of connecting people through stories — I believe in that. But I didn't know how to get unstuck. I went on a six-month search, reading all the books I could find, meeting all the mentors I could find, until I found the one that actually explained why we were stuck. Basically, we redid the entire culture of the company based on that. I redid my entire executive team. We actually built real mission, vision, values — we'd kind of had it before, but if you asked my employees what it was, they wouldn't be able to tell you.

Roland Siebelink — Host (14:42): I want to hone in a little bit on that, if you'll allow me, Andrew. As you know, we are in the business of helping to scale companies — typically 10 unicorns from one to a hundred in three, four years. And I think that self-awareness, that looking for the solution, that realization and acceptance that you need some help from the outside is so important. I was wondering — how did you get into that growth mindset versus the many founders that I see who are like, no, I wanna be protective, I don't even wanna admit to the outside world that I have any problem. We can solve everything by ourselves — and not realizing that they're stuck in a loop where no new information is coming in, so no new decision will come out.

Andrew Laffoon (15:29): Well, as I mentioned, it did take me time. I think humility — the word is humility. It's not a common word you're throwing around in Silicon Valley, like "I'm looking for founders who have humility, that's gonna be the key to returning my fund." Because humility is usually viewed as this person's a doormat, they're insecure, blah, blah, blah. I don't mean that. To me, good humility is: you have a high regard of yourself and you have a high regard for others. When you have that intersection, you end up with a person who actually can be way more self-aware. Someone who can only learn on their own and cannot learn from others — their growth rate is actually far less than someone who can learn from both. Because with that amount of humility, you can be open, you can be curious, you can ask questions. And the interesting part about humility is it's actually only possible in someone who's not insecure. Pride and insecurity are two sides of the same coin. I think as an entrepreneur, we tend to be afraid of: am I gonna fail? Am I gonna let my investors down? Am I gonna let my employees down? Am I gonna let myself down? Am I good enough?

Roland Siebelink — Host (16:41): Right. And you mentioned in the prep call how you really believe that persistence is one of the most unsung factors in long-term startup success.

Andrew Laffoon (16:50): A hundred percent. What I see is if you can just stick it out, your luck will eventually come. It might be a long time — please make sure you're working on something you actually care about.

Roland Siebelink — Host (16:56): I liked what you said about the chapters — that there are multiple changes, multiple areas of growth, typically compounding or reinforcing each other. But I do wonder: do they just happen to happen, or are they driven by circumstance? Or can you actually engineer a startup or a culture for breakthrough after breakthrough? Because I'm always thinking the only way that a sub-one-million company could grow to a hundred million plus is if you have all these compounding breakthroughs. You're never gonna get it with just one silver bullet.

Andrew Laffoon (17:43): I think I thought about this wrong earlier on. What we did — the first five years, we built Mixbook in basically an online scrapbooking business. Customers loved it, super high retention, high LTV, high AOV, and we could acquire easily. And then two things happened. One is there was no app yet on the phone for making photo books, so we said let's make a new app for making photo books. Two, one of our big competitors was copying our editor and really trying to drive us out of business because we were stealing share from them. And so I said: we're gonna pivot the whole company and go all in on this — disrupt ourselves before anyone else does. But the big mistake we made was launching it under a new brand. There were reasons — we wanted to get out of discounting, we wanted to get more press, and our space just doesn't get a lot of press. And so we wanted to free ourselves from that space narrative. It worked: we got press, we got away from discounting. But actually building two businesses at once is crazy hard. It didn't compound. With a small amount of resources, additive growth doesn't work well.

Roland Siebelink — Host (19:00): And you introduced more complexity, which is actually your main competitive advantage as a startup.

Andrew Laffoon (19:14): Exactly. What I've now learned — and had I known this earlier on, things would have looked a lot differently — I would say the two biggest breakthrough learnings for me were: one, how do you actually build culture, because that really scales well; and two, how do you turn the company into a machine that compounds. I went back and reread some classic Jim Collins work — Good to Great, Built to Last. And in Good to Great, he has this idea of the flywheel. The analogy is basically this giant wheel. To get it started spinning is really hard — you have to exert a tremendous amount of energy. But once it's spinning, it's very hard for it to slow down. It still requires a lot of energy to speed it up. And if you can figure out what your flywheel is and identify where the points of leverage are, where the bottleneck on that flywheel is at any given time, set up all the metrics so you can manage the business that way, and essentially design the organization around it — you can start to deliver compounding on a schedule. I'm making it sound a little better than it actually is because it's really a lot of hard work. But that's what we do now, and it works.

Roland Siebelink — Host (20:17): It almost sounds like you've been in one of my workshops because that's exactly what I do too. Identify the flywheel, identify the handovers, and that's the metrics — so that everyone knows what to deliver to the next function, and then just start compounding the effects of having that spin faster and faster.

Andrew Laffoon (20:50): I clearly should have gone through your accelerator 10 years ago. Probably saved myself a lot of time.

Roland Siebelink — Host (20:56): I've gone through my own learning journey since then, so I'm not sure that 10 years ago you would have had the same benefit. Excellent. So do you think that ultimately it's possible to engineer more of these breakthroughs, or is it basically always gonna be driven by events and crises and stumbling from one to the other?

Andrew Laffoon (21:16): Well, I think big companies are the example that proves you can engineer innovation. There's a lot of good examples of big companies driving incredible innovation that changes the world. They might not always commercialize it. For us, Story Mode actually came out of something we started five years ago. Because we did really well during COVID, made some money. And my philosophy has always been: venture-backed companies, you're not issuing distributions — you're putting that all back into the growth and into the mission. How can we actually make this real? And so we said, okay, we have the resources. We basically raised another funding round from our customers with no dilution. How wonderful is that? If we go back to the original vision — which was not only about photobooks, it was something way bigger — could we actually make that real? And we're like, well, it's gonna take everything we have to do that. We started that five years ago. The first two years, the technology just wasn't really ready. And still the technology's not ready for a lot of the vision, to be honest with you. Some of that stuff is still cooking a little bit on the back burner. But as soon as ChatGPT launched in 2022, it went front burner. It was: can I now use AI to tell stories for the first time? How do I give people that pen to make storytelling possible again? Initially, it was not about photobooks at all. But about a year ago, we had a breakthrough and said, my gosh, this could be commercialized right now. So we spun off a team and it took a year of refining to actually get it to be what you see today. And by the way, I was just with the team the last few days and they're like, it still sucks, Andrew. We have so much more we have to do.

Roland Siebelink — Host (22:56): When you launch it and you're not absolutely embarrassed, then you waited too long.

Andrew Laffoon (23:18): It just makes me think comfort is a negative leading indicator, always, in a company. It just shouldn't be. Work is not supposed to be comfortable.

Roland Siebelink — Host (23:22): Especially not in a startup where you're constantly fighting against the incumbents and against the dominant narrative. I love it. As a last few questions, I always want to hear more about the person behind the startup. You've been quite open about some of the personal challenges and things you've gone through in your growth journey. But what were you like as a kid, as a teenager? What predicted that you would be this successful entrepreneur one day?

Andrew Laffoon (23:54): I'm the oldest of four kids, raised in the South — small-town North Carolina. And there was not a lot of startups and engineers in that town. I was the kid who, anytime something would break in the house, I was trying to fix it. And I had a decent success rate of fixing broken stuff. But eventually I got bored and I started fixing working things. And that kind of fixing usually resulted in notable flaws post-repair. My dad was like, Andrew, I gotta get this kid something so he can fix that and not my VCR. He got me a Commodore 64. I started to learn how to code, I started to build my own software. And it started — like with many entrepreneurs and engineers — it started with hacking video games. And then my dad basically just kept introducing me to more tech, taking me out of that environment. We went to go visit Duke University and I met an entrepreneur there who mentored me and taught me, showed me the internet in very early days, and gave me a series of books on tape about entrepreneurship. There were those really formative moments. And you could also draw a through line back because I was always doing entrepreneurial things as a kid. As simple as a lemonade stand. But the most classic, which I pretty much spent entire summers doing, was going around on a golf course, collecting golf balls, washing them off, putting them in boxes that golfers would just toss. We sold repackaged golf balls on the golf course. Things like that from the very earliest days. And I still remember my first real job during high school — I was homeschooled. My parents: this kid needs time to create stuff. And school was pretty boring for me. So they homeschooled me until 12th grade, and then I was done already. I was working full time and starting a company. And I had this manager who just gave me way too much responsibility. And I loved it. Every time he would give me something, I would take it, I would finish it, I'm like, what's next? And he just kept giving me more. He never held back.

Roland Siebelink — Host (25:55): And you're so young you don't really see any risks yet either, right?

Andrew Laffoon (26:16): You don't know. And it's such good preparation for being a founder because that's what being a founder is. You just keep getting more and more and more responsibility and you have to learn how to do it, what to pass on, what to hold on to, how to be self-aware about what is now your new top priority. That's actually really good preparation.

Roland Siebelink — Host (26:37): Last question, Andrew, before we run out of time. What is the one piece of advice you would give to a younger, newer founder who is still figuring things out? Maybe they're just at their product-market fit stage and are thinking, what's next?

Andrew Laffoon (26:53): Persistence is the thing. And persistence starts with you. Managing your own psyche and your own motivation is really, really important. And the biggest thing is you just cannot do it alone. Whether that's a co-founder or a spouse or a family member or a close friend, you need somebody in your corner. And whoever that is — not everyone's gonna show up the way you think in a startup. For one thing, you're not gonna have time to be with them as much as you used to have.

Roland Siebelink — Host (27:26): Do you need somebody to just unconditionally back you, or to keep you sane, or to challenge your perspective? What do you need them for?

Andrew Laffoon (27:36): You need all of that. You definitely do not need them to unconditionally support you. I think persistence is the key. And it's: how are you as a founder getting the support you need and recognizing when you get stuck? Because that's super, super, super important — being honest about it.

Roland Siebelink — Host (27:54): Super. Excellent parting words and we'll share these with many people in the community. This has been a great interview, Andrew. Thank you so much for showing up with so much energy, so much insight. Where can people go who want to learn more about you? And is there any particular artifact you want them to download?

Andrew Laffoon (28:14): Yeah, mixbook.com is our website. Try our new thing — Story Mode. Actually, it's only in the app at this time. Download the Mixbook app, play with it — it's ChatGPT for your memories. And then for me, I'm on LinkedIn. You can just Google my name.

Roland Siebelink — Host (28:16): For people who know me and don't know Andrew yet, I'm always happy to provide an introduction as well. With that, Andrew, thank you so much — founder and CEO of Mixbook. Tremendous experience to share with our community. So grateful. Thank you so much.

Andrew Laffoon (28:41): Thank you.