打造 AI 增长最快的公司之一｜Mati Staniszewski，ElevenLabs

We met at Michael Dell’s event a few months ago, and you said to me, “Oh, I’m very curious about how starting a business today differs from how it was in the past.” Michael has over 40 years of experience and has dominated the industry for decades. When I saw him, we had a conversation about it for about 30 minutes.

You read all these biographies of the great entrepreneurs in history, and you see this over and over again. It’s like, “Oh, this time it’s different.” It turns out that no, this time is no different. And Michael says, “No, no, in fact, I think this time it is different.” So I want to start with how you think about building your company as an AI-native company today.

Yes, Michael is a legend. He must have so many interesting prospects, especially since he still runs the company. It will be interesting to see whether any of that difference applies to how the company is managed today.

But for us, in a way, we’re first-time founders. Piotr, my co-founder and my best friend of 15 years, and I started ElevenLabs. It’s the first company we started.

That’s crazy. You started in 2022. You launched before the first version of ChatGPT, right?

That’s how it was. Yes, it was still a year when the topics of the day were cryptocurrencies and the metaverse. Everyone was obsessed with those 2 topics, so 2022 was a perfect time for us because we could focus and build a lot in AI.

What did you do before founding this company?

I was at Palantir. I helped create optimization models and deliver them to clients—for example, working with the NHS during the COVID response on how to distribute vaccines across the UK, or working with the oil and gas industry to optimize energy operations. So, lots of optimization models, and then working with customers to figure out how to bring them into production.

My co-founder was at Google and was in charge of many of the text models for the Knowledge Graph. Before that, he did research at university on vision models for images. He’s an incredible mind, the smartest person I know, and he developed much of the research work. I was happy at that intersection between building the product and bringing it to customers.

Before Palantir, I was at BlackRock, creating risk models and delivering them to clients. My academic background is in mathematics. It’s a good intersection of the things I liked, and now at ElevenLabs, that’s also what I do.

From the company’s perspective, that was our main goal: Can we combine research and product deployment under the same roof? In the research area, we built all the audio models, starting with the speech-synthesis model, the text-to-speech model. That was in 2022, the first model that was finally able to achieve human quality. Over time, it has become a complete set of audio models: transcription models, localization models, and orchestration for voice conversations with agents.

Then, in terms of the product, can we unify it all into one platform that helps people and businesses transform how they communicate with their audience?

Wait, so the first idea—they started with the audio. Let’s start with the audio. Why?

1. Why ElevenLabs Started With Audio & Dubbing

The real trigger came from where I’m from, Poland. There’s something very peculiar: if you watch a film in Polish, all the voices, whether male or female, are narrated by a single person. You have a single voice narrating the entire movie, so all the emotion and intonation disappear.

It’s still like that. We grew up with it. Everyone in Poland grows up with that. It’s like one person doing the voice-over. In 2021, we confirmed that this was still happening in content.

Wait, was that still happening in the movies you were watching in Poland in 2021? The same thing that happened when you were a child?

Exactly. It’s crazy because, sure, it’s cheaper and easier, with faster turnaround times, but the quality is bad. As you can imagine, it’s a pretty terrible experience, and that was our starting point.

In the future, that experience will be completely different. You’ll have the original voice, the original emotions, and the original intonation, and you’ll really be able to enjoy it in that amazing way.

That comes at a very opportune time because just a week ago, we finally launched a model that is able to do it extremely well and carry content from one language to another. But perhaps in a broader sense, it also opened our eyes to how we interact with technology and how the way you interact with it is going to change.

That will apply to the language barrier that we are going to break, but also to the fact that general conversations with devices and with the digital world will happen differently through various modalities and channels. We wanted to build that.

Is that what you thought in 2022, or is that what you think today?

In 2022, we knew we would need to unlock stories and content across modalities and channels. We still didn’t know how quickly the change from static to interactive would happen. We expected it to happen, but we didn’t know how fast.

So, the first idea was this static, content-driven work. I’m watching a movie in Poland, and I really want to feel like I’m watching it in my native language.

Exactly.

That must have seemed like a small business at the time, right? If that was your initial idea, you didn’t come in thinking this would be a giant business. You’re one of the fastest-growing startups today.

We actually thought it was a great business back then, too. If we consider all the content and all the stories that exist, how amazing would it be if they were available in audio? Whether it’s content from one of the biggest streaming companies, television, or conversations, all of that could be delivered in the local language.

So we thought it was huge, and the truth is, it is huge. If we take this to the logical end of the conversation we’re having now, could this be the future where I speak Polish and you understand me in English, or I speak English and you understand me in any language you want?

In The Hitchhiker’s Guide to the Galaxy, there’s the idea of a Babel fish that you put in your ear, and you can understand everything around you, no matter what language they speak. We knew we’d get there, but initially, it was like dubbing.

I’ll give you the full story of how it progressed. Initially, it was dubbing, and as we began to delve deeper into what we needed to do to solve dubbing, we realized that there were 3 steps in the process. There’s the transcription step, then the translation step into another language, and then you have to regenerate it in another language.

But the research that existed at that time for each of those steps was not very good.

Which companies were conducting that research?

There were good NVIDIA models and open-source models.

But wait, was it only independent research, not some of the big companies?

No, and everything was quite robotic. You could still tell that it was a robotic voice. Nothing had really managed to overcome that uncanny valley yet.

There was a good open-source project—I’m trying to remember the repository—that was quite good but very unstable and still took a long time to generate. That was the best thing available at the time, and NVIDIA had some good research on those components, especially on the voice-to-text side.

The translation was more or less acceptable. DeepL was doing incredibly well at that time, and then you had the Google Translate version of that. But all the components for dubbing were still not good enough. So that was the first part: the research wasn’t there.

The second part was that, when we started testing the idea of dubbing with many early creators, the message we received was, “Great, I would love to have dubbing someday, but today I have other problems.”

“My problem is that I want to do post-production and change the line that was recorded incorrectly. Or, before I record my video, I want to be able to narrate the script and see how it sounds. Or, instead of me talking about a video, could I have AI talk about it?”

So, before we even think about dubbing, can you give me that? For us, it was, “Okay, before we solve the dubbing, let’s solve the research component to generate voices and make them sound great.”

On the product side, let’s deliver the ability for people to simply narrate content. Let’s slow down and ignore the language change for a second. Let’s help bring content to life with high-quality audio.

I want to talk to you about the sponsor of this podcast, Ramp. I've been reading a lot about SpaceX lately. SpaceX is one of the most valuable companies in the world, and one of the main themes in its history is constantly questioning and attacking its costs. Ramp helps many of the world's most innovative companies do exactly that. The average company that uses Ramp reduces its expenses by 5%, and one thing SpaceX has demonstrated is that a religious dedication to controlling costs can help increase revenue, because you can pursue opportunities that you otherwise couldn't. And we see that in the Ramp data as well. The medium-sized company that uses Ramp also increases its revenue by 16%. So, when you run your business with Ramp and your competitors don't, you gain a massive competitive advantage that increases over time. Ramp is the only platform designed to make your finance team faster and more satisfied. Many of the best founders and CEOs I know run their businesses with Ramp. I run my business with Ramp, and you should too. Visit ramp.com to learn how it can help your business save time, save money, and increase revenue. That's ramp.com.

Deel is how the best founders turn the world into their talent pool. I've spent a decade studying how the best founders in history operate, and one thing they all have in common is that they understand that recruiting and hiring the best talent is your top priority. Top players recognize other top players, which is why leading companies like Ramp, Shopify, Eleven Labs, Uber, and DoorDash use Deel. Many of the best founders I know have personally invested in Deel after using their product. And what they discovered is that Deel is the best company in the world building infrastructure for global contracting. Deel will help your company hire, pay, and manage any worker anywhere in the world, so you can retain top talent anywhere and dedicate the rest of your time to focusing on what you do best: delivering value to your customers. The founder of Eleven Labs has an excellent description of the value that Deel can bring to your company. He said, "We created Eleven Labs to break down language and communication barriers." "With Deel enabling us to recruit and support exceptional talent anywhere, we can accelerate our innovation and bring more voices, stories, and ideas to every corner of the world." Deel is trusted by more than 40,000 companies. Learn how they can help your business today by visiting deel.com/centra. That's from deel.com/centra.

Let me make sure I understand this correctly. You’re saying, “Let’s do dubbing. We’ll contact the creators first.” So, I’m reaching a couple of million people in English. Let me see if I can get this guy to say, “Hey, can I translate this? Can we dub this into Polish, Spanish, and all these other languages?”

And then the feedback you receive is, “Yes, that’s great, but I have many more immediate concerns. Can you help me with XYZ?”

Exactly. And it was 2 things at once, which was great, because it was also true in the research field, where you have 3 steps to work out in dubbing. Nothing in this space is good. We need to solve 1 of those steps on our own.

That’s where Piotr comes in. He’s able to create and assemble the best research team to solve it, and he himself is an incredible researcher at making that happen.

Yes, because I’m friends with MrBeast, and I remember talking to him about this a few years ago, when he was the biggest creator in the world and used to have separate YouTube channels. He told me, “Yes, well, for my Japanese versions of my videos, I don’t just dub them. I’ll hire a voice actor whose voice is famous in that country.”

Yes. Someday, you should work with us or any other company, because you should dub your content. There’s a lot of knowledge that could reach so many entrepreneurs around the world, including in Japan.

Those conversations were... It’s funny you mention Japan, because I knew we were going to talk today, and I’ve been thinking about you these last few days. I’m working on an episode about the founder of Honda for my other podcast, Founders.

The surprising thing about that is that it wasn’t Honda cars at that time, when this guy was alive. He created the most successful mass-produced motor vehicle in history, the Honda Super Cub.

The way they describe their company is, “We’re just an engine research lab.” Literally, that’s what he says: “All I do is think about engines all day.” In fact, he separated Honda R&D and made it an independent company because he thought it was very important.

They financed themselves with a percentage of sales, but transferred all the research to the manufacturing division. He says, “If you just have 1 research division and you ask them to experiment and integrate it with engineers and researchers, they’re going to constantly invent new things. Then, if you’re a manufacturer, you can find ways to apply them to your products.”

I thought, “This is—I never would have thought you’d say they’re car manufacturers.” He says, “No, this is a research team.”

2. Research + Product Deployment: How ElevenLabs Is Built

To be honest, you started with this question, and this isn’t too different from how I think about how we run ElevenLabs today.

I know. That’s why I’m telling you.

It’s very similar to a focused research laboratory. There’s research engineering to bring that to product work, and then we have many small teams working on specific product problems that we can solve. Of course, there’s the deployment and launch to the wider market to bring it to customers all over the world.

The general philosophy is to have many small teams, usually of fewer than 10 people, that have the flexibility and autonomy to move forward and apply their best judgment to what we can solve for the client.

That’s fine, but you need to say more about this because this is where I’m still a little confused. Imagine someone who doesn’t know who you are. They didn’t just meet you for the first time; they don’t know what ElevenLabs is. Describe to that person how you see your own company.

Is it like a research laboratory? Is that the word you’re going to use? How would you describe this?

He would consistently present the company as a combination of research and product deployment. The research involves frontier audio models, text-to-speech, speech-to-text, and orchestration.

The product is a platform that helps companies transform how they communicate with the world around them. That can be marketing, helping them tell stories, like Ramp creating its Super Bowl ad. It could be support, working with Deutsche Telekom to create voice agents for the call center.

Or it could be sales, helping with the qualification of incoming sales to ensure everything runs smoothly. It could even be broader operations, working with the Polish government to help create an agent who can arrange a medical appointment, follow up with the patient to remind them of the appointment, and ask how they are feeling.

That’s a combination of building cutting-edge research on all audio and, in the product, applying that audio, combining it with knowledge and creative work to enable businesses and people to change the way they communicate.

And you focus exclusively on audio?

On the research side, we’re exclusively focused on audio. In the product, we combine the best of audio with integrations, knowledge, and language models to help deliver results, for example.

Explain why you think it’s so important to focus exclusively on audio on the research side.

We believe we have incredible talent to achieve that goal. The approach is very important to effectively build the best architecture for those audio models.

We think that audio is a good combination not only of science, but also of the art that needs to be solved. It’s a bit subjective. The voices you produce will be subjective, so you really have to do it right.

But in the end, when thinking about AI models, you need data, computing power, and architecture. Regarding audio, we believe that many of the problems that still exist are on the architectural side.

We want to focus solely on solving those architectural problems so that you can truly get the best quality. Many of the people who work on our side are the best audio researchers in the world. They’re also excited by the prospect of continuing to implement the best cutting-edge audio models.

And do they feel they have technology on their side that no one else in the world has?

We believe so. That’s right.

It’s a strange analogy. I’m very glad we’re reading this book at the same time as we’re talking because, for example, Honda was constantly being pressured to diversify. He had other products. Obviously, he built a lot of things with motors.

But he said, “Does that product have a motor?” And they would tell him, “No.” He said, “Then I’m not going to build that product.” He said, “I only focus on engines.”

Sometimes it’s about 2 wheels, 4 wheels, and everything else, but he just said, “Engines.” It’s very similar to what you said about audio.

3. Focus as a Competitive Advantage

He also said that he felt he had the best engine technology in the world and that no one else could replicate what they were doing.

It’s a common question we ask ourselves internally, and it’s 100% true, how you describe it now as well. When we think about a new product, the big question is: do we believe we have a unique advantage by applying our audio models to that product experience?

If the product experience doesn’t have a major bottleneck on the audio and voice communication side, then it’s not our strength. It isn’t.

Explain a situation in which you realized you shouldn’t pursue that opportunity. Could you give me an example of what you just described?

I’ll give you 2 slightly different examples. For a long time now, and I believe it still is, the ideal approach in the audio field, if you’ve created a marketing campaign, is to combine it with a large part of the image and video work to deliver better content.

2 years ago, we would have first tried to see if we could help people effectively create lip-sync dubs or avatars for their content. If you switch from one language to another, you need to move your lips. Or, if you want to narrate something, you create an avatar.

That was about 2 years ago. Many of the models that existed in that space were simply not good enough, and launching a product there would have been such a distraction and such a big change for a company that didn’t have audio as a superpower.

The bottleneck was still the quality of the avatar and the quality of the video, because the problem to solve there is the video, not the audio. Even if you had the best audio applied to that job, the experience wouldn’t have been very good.

So we decided to pause any of those efforts. Fast-forward to today, and of course the situation is changing. I believe there’s now a good set of open-source models in that space where the combination of audio, image, and video can truly deliver that result.

So today we’re picking that up again, and the recent launch is driving us to bring it here and internationalize it, changing things up for the perfect use case for video. But the conscious decision we’re making is still not to create the model from scratch, such as text-to-video models or the open video models that exist. We want to be at that intersection where we know audio can give us the unique advantage: usually, you already have an existing asset, you need to modify that asset, add that audio component, and bring it to a successful conclusion.

Do you feel that this intense or relentless focus on audio is a defense against the larger laboratories?

Definitely.

Do they talk about this at the company?

4. Building an Ecosystem Around Voice

We do. A big part of it is that our focus is on audio on the research side, and that’s where we want to win. I think we’ve also talked about this: in the very long term, we hope the advantage we have today will still be there, but it might not be as significant in pure research. That’s why the other components are so important, that’s why the product is so important, and that’s why the ecosystem we build around that product is so important.

Perhaps to explain what I mean by ecosystem, of course there’s the brand and the trust you’ve built, but there are other parts you can build as well. In our case, this was effectively investing in a marketplace model where people can create an asset, we authenticate it, and then you can share it and earn compensation as a result. So we’re trying to create a completely different model for how that works.

Give me details about that, because I don’t know—I don’t know anything about it.

Yes. One example is Voices. We did it with Voices, where people can create a voice, we authenticate it, and then you can share your voice—your AI voice—and passively, as your voice is used, you earn compensation.

We have 20,000 voices in the marketplace today, and the people coming in now have a selection of different accents, different styles, different languages, and different ages and genders that they can choose from. Every time you use it, that person receives…

Do you, as a company, create any of these voices, or are they just third parties?

We, of course, created the system so that people could do that. In some cases, when we see that some sectors are missing, we ourselves try to search for and find people to fill those gaps and create those voices so that the wider ecosystem benefits.

Nobody knows ElevenReader. You need to do a better job. I was listening to your episode with John Coulson, whom I adore, and I don’t think I even knew about it. He kept saying, “Why doesn’t this exist?” Then, in the follow-up, he was like, “No, dude, I use that app. What you’re asking for, John, it already has. You need to show it to him on your phone while you’re recording the podcast with him.”

So I used George. That’s the voice that—

In fact, we do all these research reports for the people who come. Thank you for being a user.

That’s incredible. No, of course not. It’s a fantastic product and essentially turns every document into a podcast that I can listen to when my eyes are busy. The funny thing is, we do these research files on everyone who comes in, so I put yours in ElevenReader. I’m hearing you talk about yourself with your own product.

Well, in George’s case, is that coming from you? Is someone else getting paid when I listen to George?

Someone else is receiving a payment. He’s a great voice actor who has worked with us for a long time, and every time you hear him, he gets paid.

Understood. When we sat down, I said, “You’re one of them. I keep bumping into you everywhere,” and that’s why I said, “Dude, we have to do a podcast together.”

But you’re so confusing to me, and in a good way. It’s not a bad thing, because I was thinking, I don’t even know how to describe your company. You do all this research, but then you have all these other different products. This is very unusual, especially for someone so extremely focused.

It’s very fair, and I think we see it everywhere in AI companies, where the model becomes the platform and then the application. But for us, the unifying theme in all of them is that angle of communication. We believe that the way you communicate—how you think about content and communication—all of that is changing, and we want to be at the intersection of that.

We created the best models to help you do it, and then we built a platform that delivers that content, delivers the knowledge, and delivers the conversation in a completely new way. That’s it. Of course, there are many different applications. There’s ElevenReader, which lets you enjoy content in a completely new way.

So: research, platform, applications on that platform.

5. The Communication Platform & Deutsche Telekom

Exactly. And we explicitly don’t plan to touch anything related to intelligence, knowledge work, or programming. It is not our strength; it is not our domain. It’s a fierce battle, too—a very, very fierce battle, with many, many large companies there.

But what we would love to be, if you think about the next 3 years or however long it may be, is 3 platforms where you centralize all interactions. We want to be the leading platform for those interactions, for that communication.

Can you explain that?

The whole form of it kind of builds on that theme. The way you interact with a company’s content and data is going to change. We want to help individuals and businesses have a single place—a platform where they can configure everything, bring in their integrations, and bring their knowledge, their brand, and their business resources to implement them throughout the customer journey.

Whether it’s in a marketing example, where you tell a story through content and capture user information; in sales, when you’re trying to connect with customers and bring your product to them; by providing support throughout the entire customer journey; or simply by trying to understand every brand interaction in that process, we would love to be able to capture it and help you improve it.

Okay, let’s make this more concrete, because I still want to be able to understand it properly. Let’s take one of your customers that’s more deeply integrated. Let’s say—I don’t know, Goldman Sachs. I’m just making that up. How do they use this product and all the different products they use?

Which company, or one of the companies, is most deeply integrated with yours and can explain how they use the entire suite of products that work with your technology?

Deutsche Telekom is a great example. They use a lot of our audio work to create a podcast in their Magenta app. They also do this for their advertisements, so they can create and distribute them to their audience. They use a lot of our creative tools from ElevenLabs to achieve this.

Then they utilize our ElevenLabs agent workforce for the call center, where you call when you need help. You get that help through an integrated voice agent with Deutsche Telekom’s expertise, to ensure that if someone calls to learn about the product, they get the information quickly. If they call for support on how to get a refund or about a recent billing inquiry, they can receive that help.

More recently, they even deployed an agent inside the network. So if you’re a T-Mobile subscriber, when you call, you can ask the agent to join the call and help you schedule a reservation or translate your conversation in real time for the other person. They effectively use our combination of agent work and real-time translation to communicate in the other person’s language.

Through all of that, the information is captured so you can understand how customers interact—in this case, with the whole spectrum of the process: marketing, support, and broader operations, including sales. And we make them do all that.

Good. In a situation like that, because you obviously mentioned a big increase in your revenue—it’s growing very fast—it’s as if you’re targeting larger companies, right?

What was the first product Deutsche Telekom used from you?

The first one was marketing. It was 2 years ago. Two years ago, most of our agents weren’t really very good yet. They weren’t very reliable, and they weren’t very fast, so marketing was the most obvious choice.

They started with that and implemented it. The quality of the content was great, and people could interact with the podcast. It was the use case you mentioned: bringing in your notes and then being able to read them aloud. They simply created it daily for all customers so they could read about what’s happening in the world through a podcast with voices from ElevenLabs. That was the first use case.

Then, of course, support is a combination of voice and integrations.

We’ll talk about how you expanded them from one product to the next. How do you actually do that? They start with one, but you have 10 different products you could sell to them. How do you go from 1 to 2, and then from 2 to 5, and so on?

In that case, the main thing is how we partner with them to help make that product a reality. I think the main thing is that when you implement the first one, you make sure there’s value behind it. In any interaction with the customer, you try to make sure that we’re not just testing concepts; we test the impact, we test the value, and only after that do we try to get companies to work with us on a larger scale.

Then we tested the impact on the marketing side to get the support up and running. What you really need to do is not just the audio; you have to create the integrations, so you need to connect it with all the CRM systems that Deutsche Telekom works with.

You need to work on the integrations with the rest. How do you connect it to the phone system, with a SIP trunk link or Twilio, and how do you ensure that the connection exists? You would dedicate a lot of time to integrations, making sure that the logic is respected and that, when you answer the call, the agent behaves in the way you want.

Our 48 engineers would partner with the team and spend time building relationships, in this case in Germany, working side by side on the integration. They would make sure the agent follows the flow you want and has the knowledge it should have. The hardest thing in any voice-agent job is the actual implementation: how you test that it actually works.

You need to test it initially with simulations, verify the calls, and then implement it.

Are you doing this with full-time employees?

Yes. We do all of that with full-time employees. Then we implement it and, of course, scale over time.

When did you realize that a great way to significantly increase your income was to work with implementation engineers?

6. Forward Deployed Engineers & Lessons From Palantir

Before answering that, the work doesn't end when you implement something. You still want to continue evaluating, monitoring, and refining its behavior over time, and that applies to all use cases. Implementation engineers are also excellent at that.

I used to work at Palantir, so it always felt like a big part of how we wanted to work with clients. I almost don't like the word “clients,” because in many ways I feel that they are all partners. We work together on the same problem and try to solve it.

So much of the implementation-engineer philosophy was there from the beginning: making it specific and working side by side with partners on their problems.

This is really interesting. Tell me what you observed that worked at Palantir and made you think, “If I start my own company, I want to do that too.” Why?

I think the most incredible thing was that you were supposed to be on the same side as the client, obsessed with their problem and trying to understand it deeply from the beginning.

I used to work at BlackRock before Palantir. At BlackRock, when I joined, given the compliance requirements, for the first month or two, when you sent an email, it was verified by your team. They trained you before you sent it externally, so it was a pretty detailed workflow to make sure everything was good.

At Palantir, when I joined, after onboarding in the first month, they told me, “Okay, now you need to work with and understand the customer. You’re going to fly to Aberdeen, on the North Sea, work side by side with them, understand what’s going on, and then bring it back.”

That was a total shock to me because, given the type of approach and culture I had experienced, I had almost never encountered that with a client on the BlackRock side. Now I was there in my first few weeks, and I was supposed to go and be with them. It was crazy.

I thought it was great. I thought it was that mindset: each time, more or less, you’re going to be there on the front line, loosely speaking, to work with them and bring that knowledge back. You understand what the problem really is, what a solvable problem is, and then actually solve it.

That mindset is very true at ElevenLabs as well. All of our R&D and marketing teams are obsessed with how they can really be there with the customer, understand the problem, and work on it.

There were other companies that I thought were great, with small teams. They usually had small deployment teams working on much of that. The best-idea-wins concept was very true: within that small team, you very quickly gathered the best knowledge of what you believed should be done for that client, and you had the power to act on it.

Define small.

Around 5 people was considered a large number on the front lines, and at ElevenLabs it would also be relatively large.

I’ll tell you something more about the FDE side, because now you probably see all companies doing FDE. In our case, FDEs are part of the product team. They are not part of the marketing team. They are deeply involved in understanding what the product roadmap is.

There is a second reason for that. First, you want all FDEs to truly solve the customer’s problem and expand your product in that direction. But the second thing you want them to do is bring all that knowledge back into the product. That way, the product improves for the next generation of companies that build on it.

I feel that this second part is frequently overlooked. First, it’s fine for the FDEs to do much of the hard integration work to solve the problem, but if you don’t take the second step—learning from the experience and bringing it back into the product—then it’s basically just services.

We were talking more about that second part. You gather information through trial and error by working closely with them. Then you can take that back and spread that knowledge among the other tens of thousands of customers you have.

Yes, exactly. It’s product knowledge about how you operate with simple things. Let’s say you build an integration. How do you make it available to everyone now? That’s a simpler version, but let’s say you work in the healthcare sector. You want to optimize the product experience as you deploy it.

So, you’re using your customer base almost like R&D—another form of R&D.

You’re definitely accelerating your R&D by understanding the domain you’re working in. Problems are rarely unique to a company.

That’s correct.

The beauty of this is that, in a way, everyone benefits. You work with one client, you learn, you bring that knowledge into the product experience, and then you learn with another client and bring that back into the product experience.

Both customers benefit from having the product optimized for better performance. The domain expertise they have is still there, and they can win based on that expertise. But the product expertise—how to build based on that expertise—can be abstracted and made reusable.

7. Flat Organizations, Transparency & AI-Native Management

What else did you learn working at Palantir?

They were also a very flat organization, something we maintain at ElevenLabs. That helps with the best-idea-wins approach.

I feel like I’m imitating you with that shirt.

I thought the same thing.

I thought, “You don’t have to if you don’t want to.”

No, it’s pure coincidence. I’ll change mine.

We’ll leave that part.

There were no job titles, a relatively flat organization, and very few levels. Between me and management, where I worked before the DPO, there were 4 or 5 steps or fewer. You always felt a sense of closeness. I think there were actually 4 or 3.

At ElevenLabs, today we have a limit of 5 as the maximum depth of how many levels there should be. Hopefully, over time, there will be fewer, not more, levels if AI helps us run the organization the way we’d like.

Okay, tell me more about that. What are the things you can’t do today in the way you run your organization that you hope AI can change in the future?

You always want information from the person who works closest to the problem. Let’s say you’re developing a product: you want to talk to the engineer who actually develops that product instead of their manager. If it’s a conversation with a customer, it’s similar. You want to understand what that customer is saying directly from the person on the ground, rather than having a manager summarize that information and give it to you.

I think that flow of information will change with AI. You’ll have access to everything that’s happening with better granularity than you could ever have had before, because AI is summarizing those exchanges.

Second, at ElevenLabs, most people will have close to 10 direct reports. That’s a fairly large group, which helps us build the small-team approach that we have. That only works if you can amplify much of what’s happening across all the teams and all the people you work with, and summarize what’s happening in the teams, how they’re performing, and what some of the gaps are.

That definitely helps, and it also helps us move from reactive to proactive. In the past, you frequently depended on specific people to get information to you. Now, when you think about ElevenLabs and how we operate, you can summarize everything that happens—all the data on what works and what doesn’t—and get that signal.

I get that signal and can proactively get involved where I think something isn’t working. That helps everyone in the company, because you have that same capability wherever you are.

I believe that last part, which is closely related to the previous two, is the reason why this is possible today. We take a risk by being extremely transparent with much of the data we have and many of the documents we write, so almost everyone has access to all the company’s documents. Ultimately, that helps you create a system where you can actually leverage that knowledge.

Okay, I want you to talk more about this because I think it’s one of the topics I want to discuss with you, since you’re relatively young. This is your first company—the first company you’ve built. You jumped right into AI and were perfectly positioned for this big explosion.

Obviously, people have been talking about AI for 80 years or whatever, but you’re in the right place at the right time with, I think, the right skill set based on the company you’ve built so far.

So, how different do you think the way you run your organization will be in the future compared to how it is today?

8. Small Teams & Putting Engineers Everywhere

I believe the small-team approach will definitely be present in many of the companies that are created in the future. It also helps with a very different aspect: if you think about adopting AI technology, in our case, it doesn't have to be a top-down mandate saying, “You must use this technology to improve,” with teams then enabling others. Since the teams are relatively small and independent, people adopt what they consider best from the bottom up and can implement it immediately. That has helped a lot in that regard.

I'll give you 2 other aspects in which we firmly believe. First, in all the small teams we have, even the nontechnical teams, we incorporate engineering talent. Our talent team, operations team, and legal team all have engineers who help not only automate some of the work but also elevate everyone else in the way they use AI. I think that will be a broader pattern in the future across all companies, where more and more companies, both larger and smaller, will try to equip all teams with engineering resources so they can be smarter.

Is there a centralized function at ElevenLabs that all these small, individual teams can access? For example, I spent some time with Luca Ferrari from Bending Spoons, and he had this crazy idea—crazy to me—where he said that human resources is really important. He thinks it's ridiculous that people in technology are dismissive about it. He says, “My HR department has 50 people, and they're all engineers.”

It's like a centralized HR function that all the other companies he owns—I don't know how many there are—use. There are only 50 engineers in HR, and all the different teams and companies he owns rely on that resource.

Of course, there are centralized functions. Legal is a good example of how we manage it. It should be a very central enabler of how we help people in marketing teams and other departments.

How are they learning the trade of selling ElevenLabs?

As you said, there are so many different products that you need to be very specific about how to deliver them to a particular customer so you don't confuse them about what we offer or what we do. So, yes, there are quite a few of those functions. While they're similar to Luca's approach, they'll all have a good dose of engineering and a focus on how to scale operations so they don't become hundreds of people everywhere, but rather on how to use the few people you have and really bring that knowledge everywhere.

The most important ones for us probably revolve around RevOps—how revenue engineering works effectively. We have many tools that help you gain knowledge, whether it's during conversations, transcribing them afterward, or making it easy to fill in fields so you can capture that knowledge and not waste time manually. Of course, we're increasingly using DocuSign for much of our work, so we're trying to see how to create AI agents that help replace some of that work while operating within the team.

A good example is an AI-powered SDR. Today, when visiting the ElevenLabs website, you can fill out the dropdown form and leave information about your company, but you can also talk to a voice agent and do it in a faster, different way. They've seen a lot of people go through that process, and interestingly, 2 things happen. First, people find it easier to complete their tasks and enjoy it more. Second, they leave much more information than they would ever leave in a dropdown form. They tell us a lot about their problems and a wider range of use cases, so we can connect with them better.

That's really interesting. Since they're talking instead of writing, you get more information.

100%. It is, of course, a faster way to do it, but people also feel more comfortable than going through a manual form. I feel it's simply a better way to leave information. We think this will be such an important part that every company will have a revenue-engineering function, but in our case, we want that AI-powered SDR function to be part of the company. We want it to help everyone, no matter where they are.

We're a global company, so it's centralized and then deployed to each local team. Each local team can refine it for local nuances, but it's developed, of course, in a centralized manner.

I found one of my all-time favorite quotes while reading the book "Zero to One". The quote reads: "The most powerful pattern I've noticed is that successful people find value in unexpected places, and they do so by thinking about business from fundamental principles rather than formulas." That's exactly what AppLovin has done with its advertising platform. AppLovin connects you with over a billion new potential customers within mobile games. AppLovin lets you capture undivided attention. AppLovin ads are full-screen video ads that are viewed for an average of 35 seconds. That's a retention rate that far surpasses other advertising platforms. And you can launch on AppLovin in minutes. You set the goal and AppLovin achieves it. There's no complex setup, no experience is needed, and AppLovin scales quickly. They can put your ads in front of more than a billion potential customers. Other businesses have seen immediate results, scaling up to hundreds of thousands of dollars in daily spending and increasing their revenue by millions. So, you want to get started quickly before all your competitors are on AppLovin. And you can do it by going to applovin.com. That's applovin.com.

9. Where ElevenLabs Is Growing Fastest

Do you still think of AppLovin Labs as a single company with—how many? How many different products do you have right now?

We have 3 main product lines and 3 additional ones that we're developing.

But how many different applications, so to speak?

Within each of those, we have, let's say, 6 core applications, and within that, you can do perhaps 20 or 30 different things.

It seems like you have a lot, like ElevenReader, for example.

Yes. With ElevenReader, we have products that help with the human aspect of the cycle, to correct the content and localize it into another language.

Where does all the income come from?

The largest lines are on the agent side and the creator side. The number of use cases implementing conversational agents is skyrocketing for us today.

Are you seeing any specific industries that are adopting them faster than others?

I think the fastest one for us today is fintech. It's moving super fast.

What about Revolut?

Revolut, exactly. Klarna, Nubank—those companies are moving at another speed, as are banking customers here in the United States. Then healthcare, retail, and telecom are four of the biggest. Fintech is moving faster, of course. The regulatory aspect is a little different, but I think healthcare and telecom are kind of the next tier, and retail only just got started this year. So, they're going to move forward in 2020.

I'm going to ask you one more time. I just want to come back to this in case we missed something. Is there anything else you've learned working at Palantir that we haven't talked about that you think is valuable?

10. Taste, Art & Science in AI

This is a bit of an exaggeration, and I'm not impartial, but at Palantir there was often this concept that you need a little understanding of art to do your job well, or that people need to try to be artists even if they aren't. One of the phrases that was used was “artists' colony,” although it was never said publicly. I never fully appreciated the power of that, but now I feel it much more as we intersect with AI and the creative space in many ways.

That applies to creating audio research models—there are many new ones—and to how that work is delivered, even down to each voice that's presented.

Do you like George at ElevenLabs?

Everyone will have their own preferences, so ensuring that those voices are captured and figuring out how to meet those preferences is a complicated challenge. Second, when working with brands that are trying to define how they communicate with the world, that also requires a lot of artistry. We're trying to bring in a lot of people to combine those things.

That blend of art and science is something Palantir definitely tried to achieve. I'll let others judge how successful we were, but we'll try to do it too, and hopefully we're making good progress in that direction. Whether it's the voice market we're talking about, having lots of creatives in the company developing projects with us or around us, or having almost 40 engineers—or 40 creatives—working with clients, I think that mix will become increasingly important.

Everyone is talking about taste now, about how taste will define the cutting edge of AI. Beyond the buzzword, I think there's a lot of truth in that. More and more, everyone will be able to create everything, and what defines a good product experience or a good deployment experience will depend on how the design language sounds, how refined it is, and how you convey it.

So, I think you probably tried to do it.

We're trying to do it too.

Yes, I think what you mean is that there's a guy named Edwin Land. He was the founder of Polaroid. I discovered him because he was Steve Jobs's hero, and Steve Jobs talked about him.

I talked about that with him on the Founders episodes over and over again. I should get a framed photo of Edwin Land and put it in the studio or something. He was the one who came up with the idea that Steve Jobs used—the idea that he wanted to build a company at the intersection of technology and the liberal arts. The reason it came to mind is that, when you mentioned good taste, he has a great quote where he says, “Good taste is as rare as a unicorn.”

So, even though people talk about it all the time, it's definitely a limiting factor, and I think it always will be. But I love this idea. I think this is my problem and a big part of the reason why I don't live in San Francisco. I visit, we do a lot of recordings, and obviously I have many friends who are tech founders, but I feel that the new generation of founders lacks humanity.

Something I liked about Steve Jobs, Edwin Land, or the founder of Honda is that they repeatedly said they were only inventing technology to improve humanity. Whereas I think many of these people say, “I'm inventing technology to replace it.”

So, this brings me to something else I wanted to ask you, since you worked at Palantir. You've seen Alex Karp be one of the few people saying this, and he's been doing it for almost 1½ years. Now I really think his perspective has sunk in, where he says, “If you run these labs and keep going on TV or giving interviews saying, ‘I'm building nuclear-weapons-grade technology and it's going to take everyone's jobs,’” I see it at too many conferences, which I want to ask you about.

No, I don't mean that. Well, I was just there. I was only there for a few hours, and Karp was the first to speak and said, “What do you think is going to happen?” He says, “Your technology is going to be nationalized.” Do you have any thoughts on his perspective on that?

11. Using AI to Amplify Human Potential

Not directly. Of course, we've seen what happened with the Anthropic case recently. Since we're based in Europe and most of our team is there, we spend a lot of time with European teams and governments thinking about how we build sovereignty and how important this is going to be.

So, I don't have a direct answer to your question. I very much agree with the other part of what you said, which is that AI really needs to work for people. It needs to amplify human potential rather than replace it. I hope that we're doing a lot of the work—and we are doing all the work—in that space, but I expect that in some of the interviews you mentioned and some of the conversations at other companies, this is a growing theme about how they can bring that to life, too.

The reason I'm asking that question is because I think you and I were at a dinner with Scott Wu, the founder of Cognition. I just recorded an episode with him. I've spent a lot of time with Scott over the last few years. Even before they launched Devin, I had met him.

What I think he does well, and what you do well, is that you're always talking about the positive benefits your products are creating for people. I think you told the story of a woman who lost her voice before she got married. Then she worked with you to recreate it, so she was essentially able to say her vows in her own voice. It's a perfect example of that.

The voice is such an incredible part of identity, and the case you mentioned is similar to the work of our product, where we can work with people who lost their voices due to ALS or throat cancer and help them get them back. That's one example.

We recently worked with a musician who lost his voice and still wanted to perform, so we worked on recreating that voice. We put together a concert, and he gave a concert with his old band alongside an AI voice. Now he's touring. In the UK, he's playing venues and touring.

A congresswoman lost her voice last year and still wanted to inspire others to do the work. She was in Congress for the first time with an AI voice, trying to bring that to life, too. The common theme is that voice carries an added element of emotional impact and recognition. The moment you hear someone's voice, you recognize it if you know that person.

Of course, that has a second part, which is how we protect ourselves and think about security in a future where you can create voice. But yes, that's the work in that area. Today, we've worked with over 10,000 people to get their voices back, and we hope to continue that.

Outside the accessibility space, of course, there's what's happening in education and culture in general. Today, we're partnering with 800 organizations to help bring the technology to people.

One of the crazy stories—actually, an incredible story recently. Excuse me for interrupting. There was, first of all, a legend, a guy named Tim Green. He used to be a best-selling author and one of the best players in the NFL, and then, unfortunately, he contracted ALS and couldn't do much of that work.

Of all the things you could imagine him doing, he decided to go to the opposite extreme and created a podcast. So he started a podcast. I'm going to find this guy and look him up. No, I'm just kidding.

Yeah, he's good. You should interview him.

And he has—

That's what I meant by that.

Okay. Okay, that's fair.

Fair. Wow, I'd destroy him. No, I'm just kidding. But he does it extremely well. He has great guests. He had Harold Varner on his podcast recently, along with some NFL players. This year, he won an Emmy for his work. So it's like, “Wow.” It's incredible.

Wait, but you did his voice?

Yes, we did his voice using AI.

That's incredible.

That was one of our small, proud contributions to his work. It's so crazy. It's the most extreme thing you can do, and he does it. He's inspiring—I don't know how many people he inspired—but we had hundreds or thousands of people contact us because of him, saying, “How can we get our voices back in the same way?”

Well, I think you hit the nail on the head, which is one of the reasons—and I discovered this accidentally—that makes podcasting so powerful: the voice.

I obviously love to read. I have books scattered all over the house. I have about 1,000 books in my other library, probably 600 unread, and there are authors whose work I literally fall in love with. I read everything they've written.

But if you were to ask me about my emotional connection to my favorite authors, like Cormac McCarthy, for example, in fiction, compared to my favorite podcast, there's no comparison. Because of the voice and the human element, I feel like I know them in a way that I could never know my favorite writer.

Definitely. Even now in this conversation, the voice conveys so many more dimensions than text. With text, you imagine it and interpret it, but it doesn't have the emotion, the intonation, the imperfections, or the pauses.

Imperfection is very important. We have guests who come here and say, “I said ‘like’ too many times.” It's like, “That's how you talk.” We can edit it if you want, but I slur my words, I say “uh,” and I do all those weird things. That's just how I am.

I don't want to appear on a podcast and then have you meet me in person and feel like, “This guy didn't sound the same.” People admire imperfection—or rather, authenticity—much more than perfection.

That's true. Interestingly enough, it also applies in a nonhuman way, in the case of a voice agent. At first, we were trying to create a perfect voice agent that didn't make any mistakes. It didn't sound human.

Then, of course, the obvious thing that became clearer was the “uh,” the “ah,” and the pauses. Suddenly, the performance when working with that voice agent skyrocketed. Everyone was saying, “Wow, this is so human. This is good. I enjoy talking to it.”

So, making it imperfect is almost the key element now. The voice does convey that information. It connects you in a completely different way.

That's why I also think it's such a difficult research challenge. In text, you don't have as many of those dimensions. You need a lot of data, of course, to create a good language model, but you don't have the dimension that each voice is different and that each voice sounds distinct to each person.

Even doing performance tests for text-to-speech conversion is extremely difficult, because different models typically have different voices. That already makes them incomparable.

12. Becoming an Entrepreneur & Building With Piotr

Did you know when you were younger that you wanted to be an entrepreneur?

I'd say I didn't know this was a path to entrepreneurship.

Did you know or not?

No. I was—

Because you're European.

In Poland, there was definitely a bit of that.

I know you're joking, but I think it's true. I mean, I'm serious.

Taking the risk of starting something wasn't a common conversation at all. I was lucky enough to have an amazing family who gave me the opportunity to study abroad, and that opened my eyes to, “Okay, now you can work with some of the big companies.”

Then, when you work with those big companies, you realize you can actually do things. Palantir was great for that, definitely, where you can go and work with the client and try to solve the problem. You can take what seemed very risky to me, but you can do it.

Step by step, that opened our eyes to, “Okay, maybe this is a path. Maybe it's possible to start something of your own.”

And from the moment you realized that until you founded ElevenLabs, how long was it? Was it 1 year, 2 years?

During my time at Palantir, and my co-founder Piotr's time at Google, we started doing weekend projects together. Over the years, we tried to explore new technologies and build together.

A few years before, we knew we'd love to work on something together, but you want to work on something that you consider a real problem and that you're obsessed with. That came to us in 2021.

So, a few years—2 or 3 years?

Yes.

13. Why Mati Won’t Sell ElevenLabs

How many acquisition offers have you had?

Concrete offers? 3 or 4.

When was the last one?

The last one was last year, around June.

Are you guys still active?

No.

The reason I’m asking is that AI is changing the world. We can build the frontier of that change. We’re going all in. So don’t criticize your engineers, man. Their incentives are different from yours.

The reason I’m asking is that there are 2 reasons why this came to mind. Some of this is pure intuition; you can’t even describe it. Evan Spiegel sat in that same chair you’re in. People criticized me. They were like, “Why do you want to interview Evan?”

I said, “I don’t give a damn about their market capitalization. I don’t see companies that way.” I’m obsessed with products. That guy has soul in the game, and I hope he wins. I have no idea—I don’t know anything about their business. I just don’t like Snapchat, SPACs, anything, whatever. He’s simply different.

There’s something about him that I like. I just want him to do well, you know? I’m getting the exact same vibe from you. I thought, “Man, I want Mati to win. I want him to succeed. There’s something really likable about you.”

The second reason is that Scott Wu comes on the show every few months because I really like Scott a lot. It’s just because he’s a genius.

He’s very, very good, too.

Well, yeah, but he’s articulate, brilliant, and optimistic, and everyone’s after that guy right now. Everyone. The number of people who want to buy his company.

This is something we talk about, and I hope he holds on because I’d like to see a lot more people saying, “No, I’m here for good.” It’s not just to get a big bag of money. When you hear Travis from Uber, he made billions and billions of dollars with Uber, and he says, “That didn’t make me happy. I need something to work on.”

That’s the reason for the question, because I asked Scott that, too. He said, “Obviously, there are a lot of people trying to get him, or his whole company and all his talent, and so far I’ve just said, ‘No, I’m not going to sell.’”

Ultimately, I think they should build an independent company, too. I think they have an opportunity to be one of the hyper-AIs of the future, or whatever you call the hyper-clouds of the future. I think he can do it. They have an amazing team, and I think we can do it, too. I really mean that.

I think the opportunity that’s there right now for entrepreneurs, with this broader shift, is that things aren’t set in stone yet. It’s such a great time to build something special.

I think shows like this are so important because most podcasts are made by venture capitalists, right? The content that entrepreneurs consume is created by venture capitalists. This is kind of weird to me.

The number of founders I’ve talked to who’ve sold their companies are like, “Damn. I had something, and it was going really well. They gave me a ton of money. Now they’re telling me what to do.”

I’m like, “What did you think the money was for? Nobody’s going to give you a ton of money and let you retain the independence to do whatever you want.” It’s in all these biographies, from Ted Turner to a million other people talking about it after the fact. They made enormous fortunes, billions and billions, and they say, “I’d pay back the billions if I could get my company back.”

I was like, “Man, I think you’re smart and determined, but chances are ElevenLabs is probably the best idea you’ll ever have.” And how old are you?

31.

Okay. So you’re going to sell your best idea at 31? You have 4 decades ahead of you, maybe 5. Wait a minute. You’re going to work on your second-, third-, fourth-, or fifth-best idea? Dude, money isn’t worth it.

You’re going to get the money anyway.

Yeah, yeah, yeah. Well, you’re trying to convince me of something I’m already convinced of. I know, but everyone says this [ __ ]. All founders say this It’s really hard when someone tells you, “Dude, I’m going to give you $50 billion.”

The numbers are getting insane, so I understand why people do it. I’m not criticizing them. I’m just saying that I’m really interested in entrepreneurs for whom there is no price. Like I’ve said over and over again, it’s super important to understand this.

It’s like when everyone says, “Oh, if you love what you do, you do it for free.” No, there’s another level. If you love what you do, they couldn’t pay you to do it. How much money would you have to give Steve Jobs and say, “I’ll give you $2 billion, Steve, but you can’t work at Apple?”

He’d say, “Go to hell.” There’s no amount of money in the world that could stop him from doing it, and the world is better because he didn’t stop.

I agree with you on that. A lot of people would say that. In our case, we had several acquisition offers that we turned down, and that just wasn’t an option.

Any of those acquisitions weren’t worth $2 billion, but they would have made life very easy, of course. I don’t think money alone should be an interesting proposition. Like you said, we value that this is probably the best idea and the best time, and that everything came together like this is an incredible coincidence.

I don’t know what the world will look like in 5 or 10 years, how the conversation about universal basic income will play out, how society at large will adopt the technology, or how you’re actually going to provide for it. All those questions are there, and I think they’ll be an important part of it. Of course, the best thing we can do for the world and for ourselves is to keep building.

Did you read Scott Wu’s article in Colossus magazine?

No. My friend Patrick does.

I’ll text it to you as soon as we’re done. You can put it on ElevenLabs and listen to it on the way. Put it on an ElevenLabs player and listen to it.

His perspective at the end was something like, “Listen, he believes he’s the right person, with the right skills, at the right time.” He thinks that teaching AI to program—teaching computers to program—is one of the most interesting problems he can think of.

He was saying, “Listen, I can accept if I try and fail, but what I felt was intolerable.” I forgot the word he used, but it was something like “intolerable”—not even having tried. He was saying, “I just want to give this one shot, the best shot of my life.”

He totally understands that. I think he’s also younger than you. He says, “I’m going to give it my all and see what happens.” It’s kind of a version of how the biggest risk is not taking any risk at all. You should just go for it.

Does your co-founder feel the same way about this?

Yes. He’s totally committed, too. He doesn’t like the public eye—interviews or podcasts—but he’s a real genius. I hope he’ll come on the podcast someday, too.

He’s a great researcher, but beyond that, he can also apply a lot of those ideas to business and other areas. He understands what’s going on and how it’s happening. On the research side, the kind of AGI that’s being built and how it’s going to change the world is very much on his mind, and he knows he can be a key player.

Now the whole research team is part of that change, so it’s a unique opportunity. I think we all realize how unique this moment is and what we can do with this time and for the world.

We’ve been a company for 4 and a half years.

14. Voice as the Interface for AI

Because its scale is insane. Their view is that the next form factor isn’t a device. The way you’re actually going to interact with AI is simply through your voice. It will definitely be one of the primary ways.

Intelligence, in general, is still evolving. The next bottleneck in accessing that intelligence will be how you communicate and collaborate with it. We can solve that. We can solve that for everyone out there.

In 12 months, how do you think the way you communicate with AI will be different compared to today?

Voice is definitely part of it, but it also understands you. It’s able to connect the IQ aspect with emotional intelligence. It understands you emotionally, knows how you feel, can adjust based on that, can pause, can think, and can reinsert itself into the conversation.

In a crazy way, for decades we’ve learned how technology works and learned the language of technology: the keyboard, the screen, even programming languages. It’s like you need to learn how technology works in order to control it.

What I think we can work out now is shifting back to how we want to communicate. The most primary way is voice communication, and you can bring the technology to our terms.

I think in 12 months what will happen is similar to how we’re talking now. I think the technology will be able to understand us in an incredibly better way, both our knowledge and the emotional part of the conversation, and offer a conversation similar to the one we have now.

Yes, and if that’s the case, then the market for this—and even the use cases, the number of people using AI—will explode dramatically.

I think there’s a historical equivalent that just occurred to me as you were speaking. It’s like when Alexander Graham Bell was talking about the difference between the telegraph and the telephone. The telegraph could send messages over long distances, but you had to learn how to program it and use it. You had to learn this language to send it. He was saying that with the telephone, you just pick it up and do exactly what you already do.

Yes, 100%.

Even looking around the room, I feel like there will be so many devices that will simply be able to work on your terms.

Hopefully, the screen and the phone can stay in your back pocket because you won't need them in the same way you do now. It'll be great.

When you started the company, you mentioned The Hitchhiker's Guide to the Galaxy. Was the ideal version of your product, for you and your co-founders, the Babel Fish?

It was one of the slides where we thought, yes, the Babel Fish would exist, and we hoped to make it a reality. But not so much that we would create the Babel Fish itself, but that we would allow everyone to have the Babel Fish in their devices, in their presence, and in their current work. That was definitely one of the guiding stars we were thinking about.

15. Turning Conferences Into a Business Tool

Okay, I have another question for you. Why do you always show up at all these conferences?

I don't think I do that many. Daniel is obviously a close friend. He pushes me about this, and he's right. He says, “Dude, you have one of the most elegant businesses in the world. Everything you have to do is sit in a room, do podcasts, and everything you want in life will come to you.” His motto for this is, “Stay away from the circus.”

Yeah, yeah.

The only ones I do, obviously, are events for my partners, like Ramp, where I saw you at the Ramp dinner. Obviously, I love Michael Dell, so I go to his as well, and I don't do anything else. Because of Daniel's advice, I just stay away from the circus.

The people who show up at these things aren't doing any work; they're just there to distract. You don't have to do it.

So I get all these emails and invitations to go to these things, and I open them and see your face everywhere. What are you doing?

I usually try to do 2 to 3 per quarter, generally conferences and events. But in our case, it's a little different because a lot of the people who are usually at conferences are partners, potential partners, or clients. So it's a great way to force the situation by having them all in one place and trying to figure out how we build together.

That's the benefit of the breadth of the work we do, especially in conversational-agent work, which applies to most businesses and how they think about communicating with their customers and changing the customer journey. A lot of the events are a great way for us to find time with many of the people we work with and then convert them into customers.

Okay, so you're working on expanding from 1 to 2 or 5, whatever. I remember thinking back to the early days when we started. I got my first invitation to one of those conferences, and it was like, “Oh, this is super cool, super fun.” Then you go, your parents had also been drinking, and you feel tired. I didn't think I did anything. It seemed fun, but it wasn't very productive.

Didn't you have anything to sell at that point?

At that point, I didn't know. I didn't know what the purpose of the conference was. I think it was the lack of preparation on my part. I went to the conference and just went with the flow, and I think it was terrible.

You go there, have the sessions, and some are, of course, good, but most aren't relevant to you or your business. Or they're the circus act you mentioned, where you feel like you're doing something important, but it's not important at all.

After that, I didn't do any events, and then I realized that you can actually do them well. You do need to prepare. You need to schedule many of the one-on-one meetings you want to have during the conference well in advance. Everyone is there at that point.

You don't want to do too many because a lot of people are repeating themselves at those events, so you want to do the ones that are somewhat different. Of course, you want to be very explicit with the other party so they know what they're getting into—that you're not just selling them something while they relax. That's a recipe for disaster.

But a lot of people want that, too. They're there to do the exact same thing: do business and get things done. So now it works really well for me. I do a lot of preparation, and we schedule well in advance.

I never go and try to be there just for the sessions or just for the content or things like that. Of course, I speak frequently at those conferences, but the main thing is trying to spend time with the people who are there, and that's been working really well.

Do you still drink?

No. At weddings, but other than that, it's rare.

Every time I go to Europe, I have the same thought when I get back: I should drink more. It's so much fun. It's like they just know how to have a good time.

There are social circumstances where I would drink. They're so rare now that it's rare that I would.

Okay, next time we're at one of these rare conferences together, let's make sure we have a drink. Deal?

Deal.

Deal. Deal. Okay, buddy. Thanks for doing this. I appreciate it. Okay. I hope you enjoyed this episode. Please remember to subscribe wherever you listen and leave a review. And be sure to check out my other podcast, Founders. For almost a decade, I've obsessively read over 400 biographies of the top entrepreneurs. I'm looking for ideas from history that you can use in your work. Most of the guests you hear on this show first found me through Founders.