Lovable联合创始人兼CEO Anton Osika：实现85%第30天留存，优于ChatGPT

Thank you so much for joining me today, man.

1. How a Side Project Turned into a $200M Company

Thank you, Harry. It’s always fun to talk to you.

That’s very kind of you. I want to start pre-Lovable. I spoke to a couple of your investors in your first company, Depict, and I want to start there. What are your big takeaways from Depict that shaped how you think about Lovable?

We scaled super, super fast at Depict as well, and we liked moving very fast and scrappily. We nailed that really well. I think we also did really well on high-potential talent—quite junior, with high-potential talent. You can see that from all the companies coming out of Depict. The Depict Mafia is absolutely real.

2. Raising Series A: Should You Always Take the Money

What worked in the beginning was saying yes to a lot of opportunities and trying out what worked. Once you become more people, and you have to follow up on and maintain everything that you start, you have to be much more focused. We said yes to too many things at Depict, and we didn’t take this one thing that we could do 10 times better than anyone else. Then, as the economics and the macro environment turned worse, we didn’t continue the scaling trajectory that we were initially on.

3. Why Talent is 10x More Valuable Than Experience

I was just reading an article with Paul Buchheit, the founder of Gmail, and he said you need 3 great features in a product. It’s really that simple: make them really, really great. Do you agree with that product simplicity over feature depth?

Yes. On a product level, you should say no to as many things as possible and make it feel more like Apple. The things you do, you do with purpose.

You mentioned not saying yes to everything. Is there anything else that you did or didn’t do that really shaped how you think about the early days of Lovable?

No. I think moving fast, talent, and culture are the most important things: how you work together every day, how people interact, and how they collaborate quickly. Those are the 2 most important things for almost any company.

Let’s unpack talent, because I spoke to Fredrik at Creandum before this, and he said we had to chat about this, too. You favor talent over experience, which sounds obvious. Respectfully, you’re not going to go for an experienced but untalented person, but how do you think about your hiring lessons around experience versus talent?

Experience can be a negative thing in some cases. You often want people who are super ambitious, have a lot to prove, and are more open-minded toward how you should work together as a team.

For many roles, junior talent is first of all super easy to get into the company. They’re not already committed to many projects. The best people that you can hire at a young age will go on to become founders, and then you can’t hire them anymore. That’s why junior people are often quite good.

Will you hire someone if they haven’t done what you’re hiring them to do before? The benefit of hiring someone experienced is that you can see they’ve been at X company for 4 years and can do that at your company. Will you hire someone who hasn’t done what you’re asking them to do before?

In most cases, yes. If it’s engineering, you have to know software engineering, of course. For some roles, you definitely want someone in that domain with a lot of experience—someone who can coach and tell the more junior people what great looks like.

These are going to sound like strange questions, but you mentioned ambition there. Did you always know that you would be successful when you were younger? When you were building, did you always think, “I will be successful at something that I do”?

No, I don’t think so. I was always frustrated with how people around me didn’t understand things as quickly as I did, or at least that’s how it felt. At some point, I realized that sometimes it was me being too naïve. That’s something I learned over time.

I have a very good track record on many of the truths about what’s going to happen in the future. I think that’s one of the superpowers that has made me successful. I felt I had that superpower, but I didn’t know it would translate into building something successfully.

4. How to Use a Waitlist Pre-Launch to 10x Growth

How did you first make money? What was your first entrepreneurial thing, Anton?

I always nerded out with computers and set up LAN parties when I was a kid. I noticed that all my neighbors, friends, and family had computers with some kind of issue, and they would call me because they wanted my help. Many of them wanted to pay me afterward, so that became a bit of a side hustle when I was a young teenager.

Did you game when you were younger?

Yes.

You’re the easiest archetype. I’m sorry to be rude, but you fit all the characteristics of the successful founders I’ve had on the show: making money early and excelling at gaming. Both are very clear archetypes.

I want to move to Lovable. GPT Engineer starts as a side project. Where does the idea come from?

This was the spring after ChatGPT came out. I had been playing with the precursors to that as well, and for about a year before then I felt there was a massive wave coming from scaling up these models with more data. The path wasn’t set up for leveraging that yet.

I was traveling with my now-wife on an engagement trip, and when I’m traveling I get extra creative. I don’t think there was anyone talking about AI agents at the time, but while sitting on an airplane I started writing a lot. I thought, “You can hook them up and put the large language model in a for loop, and then you can have it do a lot of agentic things.”

When I got back to Sweden, I thought, “Where do I apply this?” Obviously, to software engineering. I had been talking to people about this, and I felt no one was sufficiently imaginative about what I was thinking. I had to prove the point that, with the first versions of the ChatGPT APIs, you could already build an agent that writes code.

I drank a lot of coffee, crammed away, and got the first version that really impressed people. You could write, “Create a Snake game,” and then get a running Snake game on your computer. That was the first version of what we’re building now.

How long did it take you to build V1 with that caffeinated session?

I think it was 1 main weekend, and then a bit of polish over a few hours here and there on 2 weekends after that.

What are the biggest lessons or pieces of advice from building many different V1s for the many founders who are listening?

It depends on what you’re building. For most first-time founders, I would really focus on the user and the user problem, and ask, “How can I get 1 person to love what I’m building in this V1?” That’s my general advice.

I didn’t do that. I just put out a video on Twitter, and it got dozens of academic references and millions of people using it.

Take me to that. We have this weekend, and you release the product. What happens then?

It wasn’t clear to me that I would build a business on this at all. I thought it was fun, and this was an open-source project, so I started nurturing a community that went on to work on the open-source project.

I went to my co-founder and said, “This thing is absolutely huge. I’ve been thinking of doing something else, to be honest.” This was a wake-up call for me that it might be time to find a good replacement for me as CTO at Depict. That’s what happened next.

5. How to Master a Public Launch: $0 - $1M ARR in a Week

A few months pass, and the community continues to grow. What happens then?

I figured out a good replacement for me and decided I wanted to have a great co-founder who could be my partner in crime here. There was a guy who was the most efficient, zero-fluff engineer and entrepreneur. He had sold a company previously, and I wanted to work with him.

I biked to his apartment and said, “Hey, let’s take a walk and plan the future.” I got him on board, and then we started building and created the first version of what became Lovable.

We’re building the first version of Lovable, and you’ve got your co-founder. Talk to me about that time. When did you release it, and how did the official release go with Lovable as a product and company?

The launch of Lovable was 1 year after we started building. In the meantime, we launched great preview versions called GPT Engineer App. That was about getting the user-feedback cycle going, building excitement about what we were working on, and employer branding as well.

The first versions were good, but they weren’t very good. We had some people really liking them, but the aha moments didn’t click for sufficiently many people. They didn’t understand how to get real value from the product.

Over the coming year, we iterated and packaged all of these things together. Today, you can ask Lovable, “I want to build a SaaS business,” and people have built entire SaaS companies and made money just by prompting our AI.

I want to break down a couple of things you said there. You mentioned the waitlist. Do you have any big lessons or advice on how to do a waitlist strategy?

Waitlists are useful because you can control exactly how many people you want to get on board and take user interviews with. Just get sufficiently many people on the waitlist, and find a good way to qualify who you want to talk to.

You probably have a different hypothesis about who you should talk to, who you’re going to sell into, and who gets the most value from your product. Qualify only those people and talk to them during the user interviews.

When you do the user interviews and feedback sessions, what are your biggest lessons or pieces of advice on how to do them well? What questions are good, what questions are bad, and what have you learned?

There are 2 different types of user research for us. In one type, we just see people use the product and observe where they understand it. That’s more of a user-experience interview.

In the other, if they’ve just tried the product a bit, we ask, “You tried the product a bit. Why are you even interested in this?” We ask them what problems they’re facing in their business and try to identify the biggest pain point they’re actually looking to solve.

It might be, “I want to get more customers, and I think I can get more customers if I can show them that I can get the first version out with AI more quickly.” Those are the types of questions we ask.

How does it change the structure of teams? Today, you have software engineers, designers, developers, frontend, and backend. How does it change the structure of teams themselves?

Harry, if you want to create a personal website, it’s super productive. You don’t need a team; it’s just you. You create it with AI. If you want to ship the first version of your SaaS and start making money, it’s all you. It’s not a team. A team just slows you down. Maybe you get some input from a designer about how it looks.

Once you have existing software with users and you want to iterate and change that software, AI might mess up your entire codebase. Then you want to work with a software engineer who knows how to maintain quality and consistently keep quality in the product.

You mentioned that people took a bit of time to find that aha moment. How important is the time to aha moment?

I think it’s very important. It’s funny, because you asked me this question and I think we’re very bad at making the time to aha moment super short. I think we could double our conversion rates if we became better at speed to aha moment.

Let me take credit for something. When you come to Lovable, you just see a prompt box. It’s very inviting. Instead of getting to a landing page, you see a prompt box, and for the people who enter a prompt, you get to a quick aha moment.

There have to be many aha moments in Lovable. It’s quite complex, and a software engineer is a very complex feature. But you’re doing that well, and that’s what I would recommend: give the user something interactive with an instant reward.

I’ve had many guests say that one of the biggest sins ChatGPT committed was making chat the default UI for the future of AI. Do you think that bluntly, chat and the prompting we have today in Lovable and many other products is the right default UI for an age of AI?

Yes. With prompting, you can do almost anything, and explaining your thoughts in written form is easy to implement and iterate on. But it’s going to get more advanced over time, with more than just prompting.

6. Why Raise a Large Seed Round

I love the way you took a deep breath there. It’s a weighty question, and you’re thinking about it a lot. You’re building the interface for creating software, and no one knows what that interface is going to look like. But the prompt remains.

I’m going chronologically through the story because it’s an amazing story. You rejected Y Combinator at some point. Why did you reject YC?

We felt that, at best, YC would mean a lot of dilution and some acceleration. At worst, it would be a distraction—to go to San Francisco and go through all of the fun things that happen when you go to YC.

We took some funding instead and built with a focus on talent.

When did the seed round come? Was that post-launch or pre-launch?

It came before we launched the first waitlist for our product.

How did that seed round go? Before you had launched the waitlist for the product, how did the round come to be?

I have advice that I’ve always followed, which is to work with investors you like. I had some people I knew from before who I thought were amazing people and who I wanted to have by my side if things went sour or if things went well.

I spoke to a few people and kept it very brief. I got a preemptive offer and said, “Yes, let’s do it.”

How big was the round?

We took $3 million, and then I got the advice to say, “Just get a lot of cash, because you never know what happens in the markets.” We raised a fairly large pre-seed round of up to almost $8 million.

Would you advise founders to raise fairly large pre-seed rounds if they can and the money is on the table? Would you say, “Take it”?

It depends on how you want to operate. If you like talking to investors—which, at the time, I didn’t; I just wanted to build the technology—the answer is, sure, take a big pre-seed round so you have time to figure things out.

If you like talking to investors, and you think it’s interesting to talk to a lot of people who care about understanding the market, then I would raise more iteratively through smaller rounds.

We’re seeing a lot of founders today be immensely dilution-sensitive from day 1 in a way they haven’t been before. Ten percent is the maximum they’re willing to give up on a round. How did you think about dilution sensitivity?

I had a very wise person tell me, “Dilution does matter so much less. It’s all about the size of the pie.” I was affected by that advice, as opposed to thinking, “Minimize dilution; this is my life’s work.” That’s mainly how I think about it now.

We’ve raised this round, and we have the waitlist. We’re skipping a little bit, so take me to go-live day. When does that happen, and how does go-live go?

We went live with Lovable on November 21 last year.

November 21 last year? That was only 4 months ago.

Yes, 4 months ago.

Isn’t it nuts how much life changes in 4 months?

It can change really fast.

So, in November 2024, you launch. From day 1, is it just nuts? How does it go?

We had paying users on an earlier version that was called something else. The launch itself wasn’t one of those “wow” launches. I think we could have gotten 10 times more press on the launch, 100%. But people started noticing that this was really good, and we continued to improve things in the product and ship things very rapidly.

Growth started ramping up after we launched. At some point, we were growing $1 million in ARR per week, and that just kept accelerating. That was in December, and it continued accelerating.

On the technical side, it was frustrating because we ran into a lot of scaling issues. The team said, “We can continue to patch this, but let’s rewrite everything while we’re seeing this explosive growth.” There were also so many quick fixes we wanted to make on the product side.

So you rewrote it as soon as possible and stabilized it?

It took a bit more than 8 weeks. It’s not completely done to date, but we spent those 8 weeks on it, and now we’re shipping faster.

7. How Sustainable is Lovable and AI Revenue

You mentioned $1 million in ARR per week. How much are you growing now each week?

$2 million in ARR per week.

$2 million a week now? It’s so funny for me, because when you live in a world of venture, you don’t get this. I don’t mean that rudely, but companies go from $1 million to $4 million in a year, and that’s a great year. Then you’re growing $2 million in ARR in a week. Does that sink in? Do you know how nuts that is?

No, I guess not. A lot of things in the last few weeks have just blunted me. I’m focused on all the things we have to fix and improve. That’s all I think about.

What are the most common ways that a company’s development process slows down? For founders listening, what should they watch out for?

Product development is usually slowed down when you have a complex product and a lot of requirements on it. You said earlier that you have 3 things that are good in your product. I think that’s generally a very wise thing to do. Simplicity leads to product direction, knowing where you’re going, and knowing what you’re working on.

Looking back since the start of Lovable, where, from a product perspective, did you invest time where, with the benefit of hindsight, you shouldn’t have?

At Lovable, we thought a lot about the community and community features inside the product. That could have made a lot of sense if we had seen slower growth, but now that growth isn’t a concern, you don’t need community features to power growth. I think that was pretty much a waste of effort.

Do you agree with the sentiment, “Build it and they will come”? The growth has been amazing from day 1. It feels like people have just come for the product because it’s been amazing and different. Do you believe “build it and they will come” stands true today?

If you have a very strong vision of where there is untapped potential, if you really know that deep down, and you have a track record of showing that, then “build it and they will come” is going to work. You also need the runway—your personal energy runway and so on—to really make it work and make them come.

8. What are Lovable’s Biggest Threats: Incumbents or Open Source

In most cases, it’s too risky to just build it and expect them to come. You can build it and make them come, or try to make them come at the same time. That’s much lower risk.

You mentioned energy, and it reminded me of a statement that someone at Revolut made to me. He said the most successful founders he invests in are between 30 and 35. They don’t have the naïveté of incredibly young founders, but they also don’t have the tiredness that older founders can have. You have more energy when you’re younger. How do you think about that, given your age today?

Energy is super important, and naïveté is often a benefit. I made a lot of mistakes as a first-time manager at Depict.

What was the biggest mistake you made?

Thinking we should change the culture and become more of a scale-up—something slower-moving, with more management layers—when we became 40 people. That was the biggest mistake I made.

Why did you think that was the right thing to do?

My co-founder and other senior people said, “Now you have to hire executives and so on.” That was a bad idea.

Did you start hiring executives, pretty much?

Yes, we did. It didn’t work. One person I hired didn’t work out, which slowed me down and set us back a lot.

When you think about that, what would your advice be to other founders?

Don’t believe the bullshit. You can scale for much longer without executives. Many founders hire more mercenary-style people—skilled people who are just running in their lane. I hire generalists and try to empower them as much as possible.

If you have a lot of generalist, super-smart people who are doing smart new initiatives, adding executives on top of them is high-risk, and the reward is questionable. At some point, of course, it makes sense.

Does culture break at any point when you’re scaling the user base and revenue so fast?

Usually, I think it does—or it changes and evolves. This is something I’m very mindful of, and it’s why I’m so scared of adding too many people.

What are you worried about?

The most important thing for everyone at our company, which I talk about, is role-modeling how much you care about the product, the users, and the team—how well the team works—and making sure other people care as much as you do.

9. How to Compete in the US from Europe

That comes from a feeling of ownership of the culture and the team. If you have a lot of people, that gets diluted. That’s what breaks, or at least what becomes harder.

We have the team scaling, the user base scaling, and the revenue scaling. We’re making a lot of money at this point. Why a Series A?

I had some scheduled check-ins with investors. There was one guy and his team who had been obsessed with Lovable for the last few months. They were hearing feedback, and he was here in Stockholm at Creandum.

He made an amazing impression on me. I thought, “I can wait, because it’s very clear to me that this is just the beginning, and I could raise later. But we can accelerate by adding an investor who is a partner in helping us find more amazing people and who can be a sounding board.”

10. Is Europe as F****** as the World Thinks

Frederick helped grow Spotify from nothing, and we decided to raise a small round.

You decided to raise a small round. I have to ask this: you have many well-funded competitors in the US. I think it was Neil Murray who said that you put Devin—which actually worked—from very early on. That was bold. When a competitor has a lot of money, do you have to raise, too?

No, I don’t think so. You could just bootstrap. You can bootstrap most things, so you never have to raise.

But they can outspend you on talent, customers, and marketing.

I’m not afraid of being outspent on any of those. The only thing that matters is execution. If you can outperform them in execution, then I’m scared of nothing.

Where could you improve your execution today? We’ve spoken about the speed of shipping, and the culture and talent are phenomenal. Where would you say you could improve from an execution perspective?

You can always improve how fast you make decisions and communicate those decisions so that everyone is really on the same page.

How do you do that today, and how could it be improved?

I think it’s about doing fewer things. We could do fewer things at Lovable as well. There are a lot of people with great ideas, and each idea in isolation is great, but you can only have so many. You should execute on only so many things at the same time.

We mentioned the team and culture, and you’ve been pretty ardent about building in Europe, keeping the team in Europe, and being a European company. A lot of people say to me, “By staying in Europe, you are deliberately not doing what’s best for your career. If you were in the Valley, you would be more successful.” What’s your response to them?

The most important thing is talent, and there’s more raw available talent in Europe. The culture in the US is more suited by default to succeeding as a startup.

What specifically do you think it is about that culture?

11. Building in Europe vs. Silicon Valley

It’s the default of thinking big, being super ambitious, and being very committed to making things work. In Europe, people are more focused on living a balanced life. In Sweden, we talk about the Law of Jante, which says that you shouldn’t think you’re better than others.

How do you respond to people who say you would be more successful if you were in the Valley?

As a founder, you have more free energy of talent to use and channel into something successful in the Valley. But there are, of course, many benefits to being there.

I think this is playing on hard mode from Europe, and I get excited about playing on hard mode and showing that you can create a category-defining company—which is Lovable, in our case—from Europe.

Are you positive about Europe moving forward? We’re in a doom loop around Europe, which we both disagree with and hate. Are you positive, and what guides your thinking?

I’m super positive. There is a strong underdog mentality among all of us founders now, which is usually a winning concept. It feels good to be the underdog and want to prove others wrong.

I also think there are incredible superpowers in using the arbitrage of incredible engineers in Europe and selling into the US. You can build huge companies selling into the US from Europe.

Being a European company doesn’t mean you can’t sell into the US.

It’s a very global market.

I’ve got some questions that are a little spicier, but I have to ask them.

Okay.

If one were to criticize Lovable, they would say, “Amazing. Look at the revenue growth and user growth, but is this AI sugar revenue? In other words, it’s not sustainable and it will churn very quickly.” How do you respond to the idea that it’s not sustainable revenue and not sticky?

We have month-one retention that’s better than ChatGPT’s month-one retention among paying customers. It’s about 85%, and that’s just going up.

Of course, there are a lot of things to be desired when working with an AI system. There’s a fraction of people who come in and flip up their credit card because they want to try it. That’s a rational thing to do, and they almost know they’re going to stop using it. We have a fraction of users who are like that.

12. The Future of Foundation Models: Who Wins

Eighty-five percent month one? Wow, that’s great. You’re also too early to have month six. What could you do today to increase retention most significantly?

The easiest thing we can do is give more users more of the important aha moments and ensure that all of our users get more of those important moments about how to use the product.

I don’t really, respectfully, get you. You give prompts under the prompt box, so you say, “Build me a SaaS app,” or “Build me a dog website.” All you need to do is click it, see the code being written, and then see it being created. The aha moment is pretty obvious.

There are many aha moments, or education moments, that help you get the full value out of using it. The most important of those happen when you feel like you’re getting stuck and the AI doesn’t understand you.

There are many things you can learn as a user to get around that. It’s about how you prompt, how you understand what doesn’t work, and how you clearly explain the problems you’re seeing and what you think could be the problem.

When you’re building a more complex feature, you can also onboard an engineer to make small changes to the codebase. Those are things our users should know, and not everyone knows them.

What’s your North Star metric today? If you have 1 metric on the TV for the whole team to focus on, what is it?

It’s the number of users who go all the way to getting users on what they built—getting something hosted with users on it. That’s what we focus on.

What’s the amount today?

We have almost 40,000 paying users, and that’s the proxy for it.

Forty thousand paying users is incredible. Do you care about the time it takes to go from starting to having a website created?

We care, but I have to say, Harry, the onboarding and how much it can improve haven’t been a focus for us. We’ve just been making the core AI parts better and better. That’s what we focus on.

People also say, “Lovable is a wrapper on top of other people’s models.” Why are they wrong in that assumption?

It’s very easy to create a cool demo with just a wrapper. It’s super easy to make a cool demo with just a wrapper. The hard part is getting close to 100% of what you’re asking for.

There are a lot of small details in making that work. There’s a chain of large-language-model API calls and other algorithms that run, and that chain is something you can continue to optimize for years without reaching perfection.

Whose models do you sit on top of today?

We use all of OpenAI’s models, Google Gemini, and the main workhorse for writing code is Anthropic’s Claude model.

13. Grok vs OpenAI vs Anthropic: Buy and Short

I was going to have this in a quick fire, but I have to ask it now. You’ve got Anthropic at $60 billion, OpenAI at $300 billion, and Grok at $50 billion. Which do you buy, and which do you sell?

I care about the best talent here. I feel like Elon is very good at talent, so I would buy Grok. I think they’re also very ruthless at finding business opportunities. We’ll see about that, but I feel like they could be good at it.

Anthropic is my favorite. I love the culture and the leadership there. I would short OpenAI because they’ve been very good in the scrappy phase, but they haven’t proven that they can have clear product direction and focus over the last few years.

I’d push back on you. The 2 biggest things that matter in this next wave are brand recognition on the consumer side and consumer-facing frontend products.

When you look at brand, everyone’s mother knows ChatGPT. They don’t even know OpenAI, but they know ChatGPT. On the consumer-product side, respectfully, they’re way ahead of anyone else. Am I wrong?

That’s true. I just think we’re still early in the days of AI. If you look at enterprise revenue, Anthropic has almost caught up to OpenAI from nothing, from being absolutely dominated.

I don’t know what xAI is going to do here, but I expect them to be able to pull something off. OpenAI lost all of its best talent to Anthropic, and I think xAI might be able to pull something off here.

I spoke to a very wise friend of mine before this, and they said their biggest concern is that the open-source models or the megacorps of the world, with massive distribution advantages, come in and win the market. How do you think about that, and is it a concern?

The big megacorps move very slowly in many domains, so for many years they’re not going to have the best product on the market. There are distribution advantages for some of these megacorps, and that’s what I’m more concerned about: the marketing and distribution.

Overall, it’s going to be a growing market where a startup can easily be best-positioned for some parts of it.

One real shift you’re going to have to face, I imagine, is the shift from PLG and prosumer to enterprise. How do you think about that shift, and is it not a concern given the speed of your revenue ramp?

If we did enterprise, I would want to do it really well, so we’re holding off on enterprise for now. Our goal—you asked about the North Star metric—is to be the best place for builders to create products and to get 1 million of the most talented builders on Lovable.

If we succeed in that, it’s a great segue into many other areas, including enterprise.

When you look at Shopify, what do you learn from the trail they’ve blazed in the way they’ve executed so efficiently?

I’d love to hear from you, Harry. What are you most impressed by with Shopify?

Honestly, I’d say it’s their narrative around everyone being an entrepreneur. Harley is a phenomenal communicator and storyteller. He brings very real businesses to life and explains how Shopify changed their processes and sales and really made an impact.

I think he’s obviously a brilliant technologist, but without Harley’s storytelling and his ability to bring entrepreneurial vibes to life, it would be a different company. I would learn a lot from that in terms of how to tell your customers’ stories.

Not totally. I think Shopify has been able to execute on many things and create a good package. They’ve iterated fast on serving all the needs of building your e-commerce business. That velocity was one of the most important things in building the best product.

Looking forward, what would you most like to build into the product but, for whatever reason, cannot? What would that reason be?

I think I can build anything into the product, really. The team would block me. It’s prioritization and the fact that it’s not the right time.

So it’s something you have to say no to right now?

I’d love to build a way for founders to get everything you get out of Y Combinator into Lovable. That includes marketing infrastructure. Y Combinator has the entire playbook, and we could go beyond that and say, “Here’s your incorporated Delaware C-corp company and your Stripe setup. You just have to focus on the product, talking to users, and creating content for marketing.”

When you look at the partnerships and APIs we have with the different providers today, that seems like a real and possible product.

Yes, it’s on the roadmap.

Y Combinator is going to love you after this show.

Looking forward, what concerns you most? If you had to choose, would it be regulatory challenges, competition, or hype cycles?

If a competitor were super, super good at marketing, that would concern me.

How important is brand?

Brand is mainly the outcome of your product. There are other factors, too, but product is the important thing. A good product plus awareness creates brand. That’s sufficient.

So it’s more of a downstream effect of something else.

It is important, though.

You’re doing $2 million a week in additional ARR, so more and more investors are going to want to give you money. How do you think about that? What makes it worth taking, versus thinking, “This is a distraction; get out of my face”?

Currently, it’s a distraction. It might become worthwhile when we know exactly how we want to spend the money, or if it’s a partner we really want to work with, on terms where we would never regret saying yes to them.

What, if done, would be a massive needle-mover for the company?

1 or 2 more perfect technical product hires.

14. Quick-Fire Round

I could talk to you all day, but I want to move into a quick fire round. I’ll give you a short statement, and you give me your thought.

Yes.

What do you believe that most people around you disbelieve?

I think the models we have today are really smart. That’s where people don’t agree with me. They’re smarter than humans, but they don’t have memory to the same extent we do, and they don’t have context.

How fast will we ramp context and memory, given the scaling today?

The thing you need to ramp is deciding on the process for storing memory. That’s what you have to ramp. The system has to store all of these things somehow. My conversation with Harry has to be stored somehow, and almost my entire childhood has to be stored as well. I don’t know; it’s going to take years.

You can buy and hold 1 public stock for the next 10 years. Which one do you buy and hold?

Some talent play comes to mind, and Tesla is interestingly positioned outside of software.

What was the most important trait in a founder that no one talks about?

Great judgment is something people probably talk about, but I like people who see potential in others. They think, “This person can become amazing.” The judgment involved in picking talent like that is important.

What’s your biggest weakness as CEO today?

I’m not as good at multitasking as I would like to be.

If Lovable failed tomorrow, what would be the reason? Investors write a premortem, which is a reason why something doesn’t work. What would yours be?

We lose momentum and excitement. That’s what fuels us today—the excitement and the momentum.

What have you changed your mind on in the last 12 months?

You don’t need to be attached to 1 foundation-model provider. They’re all going to be amazing. There isn’t going to be 1 winner here.

Are we seeing foundation models become completely commoditized?

Yes. We won’t see specializations like we do today. I know Claude is better for coding and engineering, but I think that will equalize.

I love the Swedes. One-word answers.

What’s your favorite failure?

When I was 17, I failed my exams at school badly. I got Ds across the board. I looked at myself and thought, “I’m either going to work behind a bar, or I can make something of my life.” I chose to work harder than ever and make something of my life. That’s my favorite failure.

Probably that Depict didn’t work out, so I could do more things.

What concerns you most in the world today?

Leadership in the world today. I want leadership to be true idealists, not a function of circumstances or people-pleasing. Leaders should be much more open-minded toward people who disagree with them, especially when those people are super-intelligent.

Leadership in the world today is usually a bit corrupted and narrow-minded, and that’s what concerns me.

Do you think they would say you’re naïve?

I want to be naïve about the challenges of leadership and the challenges of having a voter population. Sometimes you have to behave in certain ways.

I think we used to have better presidents and better leadership throughout the last century. I don’t think I’m too naïve, but I’m definitely always a bit naïve.

I totally agree with you. I think the state of leadership today is woeful.

My penultimate one: what’s your biggest short in the public markets?

I’m not very well-read on the public markets. It would be a SaaS company with per-seat pricing where its ICs are going to be replaced by AI that simply replaces the employees. The number of seats goes down.

What’s the future of pricing in an AI world?

It depends on how defensible your business is. If you’re selling to enterprises that never change their SaaS software, you should try to use value-based pricing. I don’t know how you get to good value-based pricing, but I would do something like that.

Otherwise, I wouldn’t innovate too much on pricing. Do whatever works today.

I love the facial expressions.

My final one: I like optimism, and I think we need more of it in the world. I’m incredibly excited for the next decade because I think we’ll find cures for diseases that we really didn’t think were possible to cure before. My mother has multiple sclerosis, which is a horrible disease.

What excites you most about the next decade?

I think it’s that AI will hopefully make us humans understand each other better and become better at playing win-win. That would be awesome. I’m excited about that.

Our leadership actually being enhanced by superintelligent AI.

Anton, I’ve so enjoyed doing this. Thank you so much for joining me, and thank you for inspiring a generation of European entrepreneurs that they can build unbelievable businesses in Europe.

Thank you, Harry. It was super fun.