ElevenLabs：打造AI销售机器，以及我们为何设定20x销售配额

Carles, listen, I know a show is successful when I get friends of mine outside of tech saying, “Oh, loved the clip with Carles.” And I’m like, “Really?” And they’re like, “I never thought that quota and sales comp would be so interesting.” Then I’m like, “Thank you, Sarah. I had no idea you were interested in it.”

It was such a successful show, I thought we had to do a round 2. Thank you for joining me, man.

Thank you for inviting me. This is great. I’m happy people were interested in sales and sales comp, according to that clip, which now has 5 million-plus views, which is insane.

1. The new CRO playbook: how go-to-market has fundamentally changed

Dude, I actually wanted to start with this: you have now kind of led the way in a new CRO charge, I think. ElevenLabs is probably one of the most exciting companies on the planet. You’re a CRO there, leading the way. Are a whole generation of CROs being left behind?

Some of them, yes. But I think it’s fundamentally that the market has changed. The way that we build companies today has completely evolved. We can do deals much quicker, but for me, the key item is that a deal can be done. You can hire teams that have a ton of experience, but how do you think beyond a single deal and think about distribution entirely, right?

I think that’s what we’re actually missing in many ways. Yes, it’s great to do business development. It’s great to think about affiliates. How do you think about it much more on a strategic level, but also be able to execute? For me, those are the key components. How do you embed AI in your entire distribution component so that you’re able to actually do more with less, right?

What do you mean by that? Is that simply using Artisan, Qualified, and Monaco, and having AI SDRs do outbound for you?

2. Why AI outbound tools don't work — and what ElevenLabs built instead

No. That doesn’t work.

Good. Glad I suggested that.

It doesn’t work like that.

So you’re like, “You’re not hiring me,” huh? Dodgy start to the interview.

No, you’re fired. No, but I’ve tried a fairly large number of AI go-to-market tools, and they don’t work. They don’t work because they see everything as a transaction.

If I was building an AI agentic platform for sales, I would say, “Okay, let’s actually understand each one of the core potential leads. What do they prefer? Do they prefer to be reached out to over email? Do they prefer to attend events? Do they prefer to have a phone call?” You can try to map some of those components, but what people put on LinkedIn and social media is the most fundamental element.

What we’re seeing on the AI SDR front is that the majority of these tools track everyone as if they want to receive a message. People hate it. The response rates on outbound emails have dropped to the lowest point at any time. It’s less than 0.01%. The number of people getting messages on LinkedIn has skyrocketed. You can perceive that it’s a transactional email mass-sent to everyone, or that it was generated by AI. That doesn’t work.

So, outbound is dead?

Outbound is dead unless you do it with humans, or unless we do it humanly. That’s fundamentally the key item.

I can tell you that at ElevenLabs, I spent the past 2 years-plus trying to convince the team to hire engineers to help me build AI agents for revenue and for sales. It was only last year that they were like, “Okay, let’s do it.”

We ended up hiring people and building an AI SDR that handles inbound with super-good results. I now have an AI proposals manager that scans the web for RFPs and RFIs, then scores and proposes things. I have an AI customer success manager that fundamentally lives in the background of email and essentially proposes emails.

It looks at all of the data of a single customer, all of the pricing tiers, and everything that we have in the contract, and then proactively proposes things. When my AI customer success manager goes in the morning to check the emails, that person has a number of drafts that the AI has created. Then that person can change those drafts, put them out there, and send them back.

We store what we actually sent, what it was originally created as, and what the responses are, to actually fine-tune. Each customer ends up getting a slightly different message with a slightly different tone. If they speak other languages, then it’s a completely different language.

That works, and that has closed deals for us already. We generate money from that AI customer success manager, fundamentally because it is human-like, right? It’s human if it’s perceived as human.

3. Commission structures: ElevenLabs' 20x quota model explained

Totally. That’s an impact, for sure. I totally get that.

4. Will AI agents shrink sales teams? Carles' 50% productivity goal

So, we’re cutting the size of our team. If we can do so much with these different AI agents, our sales teams of the future are going to be dramatically smaller. I think we will end up seeing that people are becoming much more efficient.

I think the goal for me at ElevenLabs is a 50% improvement in productivity. That’s what I want to do, because then it means that I can hire fewer people, right? That’s fundamental.

I prefer to manage a smaller team that works super well and is compensated extremely well. Any upsells or contracts that the AI agent ends up closing, I will pay those commissions as if a human actually closed them. I’m very happy with that.

But it also means that the people we want to retain are going to be extremely top talent, performing at ElevenLabs at a level that no one else is performing at. That has a consequence. Instead of doubling my team, doing a 5x, or doing whatever it is over the next 2 years, I end up hiring fewer people. But they’re very concentrated.

Totally. You mentioned commissions there, and I’m really intrigued. We see that OpenAI doesn’t have commissions. It’s like, “Here’s your salary, and then here’s your equity.” Often, people say salespeople are coin-operated.

You said last time that you had a 20x quota, which is a lot higher than anyone else. We recently interviewed the head of sales at Clay, and they said there’s a 6–8x quota. Post-20x, what is the commission structure, and how do you advise me as a founder on how to set the commission structure?

I can tell you that we had 2 employees who, in February, had already achieved the entire full-year quota. I put a message in Slack saying, “These 2 people have reached Mount Olympus of sales at ElevenLabs,” because in 2 months they had already done their quota.

Did you not massively miss that quota, then?

I don’t think so. I think you need to put a quota that is challenging but also fair. If people miss it fundamentally because it’s too high, then you need to do the right thing and lower it or compensate them correctly, right?

But good salespeople want to be the best ones. They’re moved by the coin. They’re moved by the fact that there’s a big challenge ahead of them. If you don’t put that challenge, they’re just going to be slacking. You’re not going to get the best out of them, right?

For me, that’s the key item. We say, “Hey, you reach your quota at 100%—fantastic—and then you start unlocking accelerators.” I’m very happy if we sign a $1 million commission check. I’m the happiest person alive, fundamentally because that person did so much money for the business, for our shareholders, and for the rest of the company. That is absolutely a no-brainer.

For me, for every $1 million in revenue that any single person signs—and I don’t care if it comes from the SDR, the account executive, the CSM, or an engineer who went all in and managed to get a contract—every single million has $33 million in extra valuation for the company.

If you put it that way, it’s a benefit for everything. Some people get more equity on the engineering side. Great, they deserve it because they don’t get commissions. Fantastic. They were propping up the entire value of their stock.

When you get to the accelerators, how do you advise me on how to structure them?

It depends. The way we do it, and I think it’s fair, is that for every single extra tier, you get another 25% extra commission on top of it. We start at 5% commission on anything that you sell, and then we have the accelerators layered on top.

After your quota, that is.

No, that is including your quota. After your quota, yes, we have additional accelerators. We have a 1.1X, 1.2X, 1.3X, 1.5X, and so on. The more you keep selling above your quota, the more of those accelerators you hit.

We also have accelerators, or spiffs, for individual products. If we are incentivizing one product because we want to incentivize it in a given quarter, then we will put an accelerator or a specific spiff on that product, and people become even more motivated on that front.

Do you find that always works? Accelerators or not?

No. I think it's easy to get hooked on spiffs or specific incentives and lose sight of what you're trying to do. If you put too many accelerators or too many incentives in one place, you create the wrong behavior, which is people trying to sell whatever it is to unlock the incentive.

For instance, we don't pay commissions on pilots.

No.

Why? Because it's not adding to our valuation as a company. If it's not adding to our valuation as a company, the engineers, researchers, operations people, and everyone else aren't getting a boost in their equity. So why should we pay it?

Let's do the right thing, and let's do the complex thing. Let's do the pilot if needed, convince the customer that the metrics are there, and do it really well. Then, once we sign a yearly contract or a 2-year contract, we pay you the commission.

So you pay on retention and expansion.

Yes. In the first 12 months, you can close a deal. It doesn't really matter whether it's $50,000, $1 million, or $10 million, whatever you want. Anything that gets expanded over the next 12 months earns you commissions. You can hit accelerators with that as well.

5. Hunters vs farmers, customer success as a revenue function

If it's a strategic account, and we have a designation for what a strategic account is—for instance, the top 20 or 30 accounts in a given market, depending on how big the market is—then fundamentally, you unlock commission for 2 years. We can change it if needed, but at a practical level, we found that is the fairest way.

I will never forget one of the great CROs saying to me on the show, “You do not ever want your farmer going against someone else's hunter.” Meaning your kind of cushy but super-nice customer success enablement—“Hey, let's get champions in your company”—kind of happy, against someone else's all-star hunter who is there to close your business. How do you feel about that? Is that true? And how do you think about retaining the hunter on the account?

The hunters are important. Very important. Specifically, when you think about distribution, outbound, inbound, and a lot of different pieces, they are important.

At the same time, a hunter who is not well managed will create more damage than opportunities. Let's say, for instance, you're trying to close a company that has multiple companies or multiple subsidiaries. It's a big holding company—we can use any name you want—and you have a hunter who essentially tries to go to each one of them but doesn't do it in a very structured way, trying to align the incentives for this big account.

Fundamentally, you end up having discrepancies in pricing and discrepancies in the actual value for them. Each one of those discrepancies ends up coming back to hunt you down when you try to renew the contract or the order forms. It is important to have hunters, but at the same time, you need to keep them in check together with customer success.

I am a big believer in customer success. I don't believe customer success is a satisfaction or happiness moment for customers. Customer success needs to be a money-generation function for the business that works for the customer but also works for the business to incentivize growth, expansion, and cross-sell.

What do you mean by that? I had Chris Dagnall on the show, one of the all-time great CROs at Snowflake—zero to 4 billion, or whatever it was—and he says customer success is complete bullshit. You have professional services; pay for it and we'll make you great. To what extent do you agree, or do you think customer success is a growth engine?

In the world where Snowflake grew up, that made sense. In the world we are in today with AI, that doesn't make any sense, because fundamentally, anyone can spin up a competitor to your product in the next 2 days. Anyone.

Your customer success team needs to go as deep as it can as early as possible and try to close a single contract as soon as possible. Then your customer success team is the one that will actually help you expand it quickly and retain the customer.

If it becomes a services business only, where you're charging for every single penny, then you become a transaction. You're not building a community, and you're not retaining the customer for the long term. You should be incentivizing them to do both things: build community and trust, and also incentivize the long term. That's why I am in favor of charging for services, but not in all cases does it make sense, fundamentally.

6. The myth of "one market at a time" & hiring experienced sellers

Totally. The game has changed. Where else has the game changed? Where else are old-school CROs out of date?

In general, going one market at a time doesn't work. The traditional method has been, “Oh, you open one market, you go deep, you win the account, you show the value, then you go to the next one, and then you go to the next one, and so on.” This is the advice VCs have been giving for many years.

I know it's changing fundamentally because the reality is, if you've got 100 competitors in the next month because you're starting to get some traction, or because of whatever reason, or everyone is thinking about the same problem, then opening one market after the other instead of parallelizing and trying to make multiple bets at the same time is just counterintuitive.

Is that only enabled by a PLG motion, though? A PLG motion allows you to do multiple markets at the same time much more easily, because if you have an enterprise motion, you need everything from SDRs to AEs to customer success in that place.

No, not really. I know this is going to be controversial, but I think all of the tech startups are afraid to hire someone who has had 20 or 25 years of experience. The excuse they use is that the person is not going to fit the business, and I think it's completely wrong.

Fundamentally, someone who has 20 years of experience in sales and has been selling for the past 10 years to the same financial services firms in the City or in the US has so much wealth and so much knowledge that they can propel the entire business.

The question is that not all of them are going to fit your business, but a good portion of them, if they're still hungry and still want to do something meaningful, are going to fit really well. Those types of profiles will shorten your sales cycles because they will join and be like, “Okay, now I know the product really well. I know the vision. I know how to pitch it properly. Let me call all of my friends and all of my stakeholders that I've been interacting with for the past 10 or 20 years, and I will just go directly to the C-level.”

Dude, is that not a massive misalignment of cultures when you think about the super-young, hungry reps, and then you bring in 50-year-old Simon, who's been selling into City financial services for years?

The biggest example is Google in the early days. You had Sergey Brin and Larry Page together with Eric Schmidt. That's amazing. A guy who had 20 years of experience, was really successful, ended up coming in, and magic happened.

Fundamentally, if both parties want to make it work, it can happen. But you need to make it work, and you need to hire the right people who will be on the same brainwave as you are.

Do you worry that people pull ElevenLabs so much out of your hands that anyone could sell it?

No, I don't think so. I don't think it's that easy to sell, especially in this environment. Of course, if you have a big brand like OpenAI or Anthropic, it's a little easier.

You can't have a 20X quota and say it's not easy to sell. You can't have people who are that much better than everyone else.

I think you can.

Your people are not—I love you, dude—but your people are not 4X better than everyone else in the market.

I do 100% believe that our people are 4X better, without thinking twice. I would take them to war. I would go to war with every single one of them.

It's because we're tough on them, but they're tough on us. They give us the difficult feedback, and I get it every single day. I'm happy that they give me the tough feedback.

What we're trying to do here—and I've always believed and grown up with the idea—is that we're trying to find the ground truth. What ground truth means is that today it means X, tomorrow it means something else, and the following day it will mean challenging you and constantly thinking beyond what is actually happening today.

You're not going to be able to challenge yourself when you don't have an answer. Which market should we be opening? How deep should we go into some of the structural changes? The company is growing very quickly.

Do we need a global accounts team? I’m already thinking about a global accounts team. When do we implement it? How do we implement it? How do we motivate people? Should it be 20x? Should it be something else? What product should we be selling? Should we be selling to our competitors? Should we be selling to resellers?

There are all of those things. The team is challenged every single day, and I love it because it’s intellectually stimulating. It also forces the entire business to rethink our position every single day.

So you would advise founders today to go to as many markets as possible as soon as possible?

Not as many markets as possible. I think you need to map out and have a reasoning and a thesis for why that market, so that you can always refer to that and understand whether the market conditions have changed. Then you have to evolve your thesis.

For me, go-to-market is similar to investing in venture capital, right? You need to test 100 things to actually find the 3, 4, 5, or 6 things that perform and do really well. We do the same in venture, right? We invest in companies knowing and believing that all of them are going to work and that all of them can be a billion- or trillion-dollar company. Then we realize, “Shit, it didn’t work, but I still have 3, 4, or 5 that are going to work really well.”

7. What didn't work: failed bets, the India reset & going back to zero

Go-to-market is exactly the same thing. Test as many things as you can—not only in opening markets, but also whether you need to try self-service, work with resellers, partners, the cloud ecosystem, a grants program, or an affiliates program. There are so many things, and opening markets is one of them.

What did you test that didn’t work, and what did you learn?

At ElevenLabs, I’ve tested endless amounts of things, and I absolutely love it. For instance, media and entertainment did not work for us in the beginning. It just didn’t.

What does media and entertainment mean?

Big brands, like big entertainment studios. The thesis was that they were generating so much content, so we should be able to sell to them and make a ton of money from it.

The reality was that, yes, they do make a lot of content. But when you have an industry that is heavily influenced by the public and the audience, and by the quality that you need to produce, events, and all of that stuff, then potentially the industry might not be ready for the technology that you’re trying to sell. That was a big learning.

We spent months and months trying to break into the entertainment industry, only to realize that we were not growing as fast as we wanted. So we ended up switching and having another ICP. It was, “Okay, let’s work with media creation platforms.” That worked really well for us.

Then we ended up saying, “Hey, actually, the world is moving. We had a thesis around the world moving towards agentic systems. How do we also bring a product to market that fits really well in the agentic world?” We had a vision for that, so we ended up working on it.

Today, AI agents are actually fantastically huge for us. Fundamentally, that’s because we had the vision, we bet on it, we went early, and we started working with and partnering with some companies. That panned out really well.

You said you weren’t growing as fast as you wanted. Now you’re an operator on the other side of the table. You’ve seen how fast a company can grow with ElevenLabs. You’ve also seen it with Revolut, to be fair. To educate us, you need those two educations.

That is the business of venture. My favorite thing with VCs is when you get other VCs saying, “I don’t know if Carles is very good, because if you actually take out ElevenLabs and Revolut, he’s not that good.”

And you’re like, “Yeah. One of them is a unicorn, so that’s fine.”

Yeah, you know, okay, sure.

But my point is, has it changed how you think about attractiveness in company growth?

I don’t know. I think you still need to have companies that grow from 1 to 4, or from 0 to 1, and it takes them 18 months to get there. I think that’s actually not bad. It means you need to incentivize the portfolio to be able to iterate and experiment as much as they can, because we don’t know what’s going to work.

It’s the same for us. Even with new products that I’m heading and trying to get to market, I truly don’t know how we’re going to price them. I truly don’t know if they’re going to work out. I’m just trying to iterate. Is it going to take us a little longer to get to the first million, the first 5 million, and the first 10 million? Fine. We have the ability, but we need to iterate.

At ElevenLabs, we have a business that grows extremely quickly. We’re smashing it, all of that stuff. But at the same time, we have brand-new bets that are not generating any revenue. That’s great, because they’re bets we’re making for next year.

What big bet have you got today that’s not generating revenue?

We’re generating a good amount of revenue, but government work, fundamentally, is just a tiny portion of it. I think that’s okay, because the role of a CRO is fundamentally thinking about not the revenues today, but the revenues of tomorrow.

I’m already thinking—this entire quarter, I’ve been thinking—about how we put the pieces together so that ElevenLabs next year grows even faster than this year. I know this year we’ll smash the numbers. I have the right team, we’re going to continue hiring, and we have the motion that we set last year. It’s going to grow really well.

How am I doing it for next year, being very creative, so that I can start putting the right bets? That’s what my quarter has been this year.

You’ll hit a billion in revenue by the end of the year?

Who knows? I would love to hit a billion dollars in revenue by the end of the year. We have some targets internally. If we’re creative, we can do it. If not, it will be a little bit later. That’s fine. That’s absolutely fine.

Any big lessons on forecasting?

Hard to do. It’s just impossible.

But that’s different. Traditional sales leaders on the sales floor for the last 3 or 4 years—because I’ve done 20VC—I’ve been like, “Oh, I can pretty much get there to 3% to 5%.”

Yes. And I’m like, “Wow, now I have no clue.” But it’s impossible if you think about it from an experimentation perspective. I always tell my team, “I want us to test as many things as possible.” I only need 1 of those things to work really well to give me another 100 million dollars in revenue, or 200 million, or 300 million.

When you say things, do you mean channels, like affiliates, partners, and referrals, or do you mean products?

Channels, products, ideas. Should we be pitching services?

I was in a meeting with the CEO of an airline a few days back, and we were talking about agents and all of that stuff. At some point, he said, “But Carles, everyone keeps pitching me. I get over 100 pitches every week from companies doing customer support. I mean, I don’t want to have another pitch on that.”

I said, “You’re 100% right.” I hated it. Actually, don’t get me wrong, we do it and we do it really well. But at the same time, it’s boring. The majority of companies will start with bottom-line optimization.

So what’s boring? Customer support?

Customer support optimization, for me, is like—yes, your mindset is, “Yeah, I want to automate and increase my profitability by a bunch of percentage points. So that’s what I’m focusing on.” And that’s great. Don’t get me wrong, it is fantastic.

What I’m actually interested in is how we figure out a way to partner with a company and create a labs area. So it’s Company X Labs, right? That way, we can figure out a way for them to generate new revenue opportunities.

That’s what actually makes me passionate. But we have to prove that that is the right thing. That is one of the experiments, and I’m spending a lot of time these days on how we prove that the idea of labs for companies—where we essentially are their consultants and help them build those agents that will generate top-line revenue with new ideas and new products—is something that we can scale and use to generate hundreds of millions.

How do you think about the trade-off between doing that, where it’s maybe a little bit lower margin given the fact that it’s quite hands-on—I’m thinking British Airways Labs, where you have amazing voice agents who call you up and are very personalized, giving you tips on how to enjoy New York with your wife—and actually doing avatars, where we can compete with all the avatar companies? The former may get more money from it. Boom. The latter has higher margins and is a very adjacent product.

I don’t know. I think I would rather do things that are actually much more complex than the ones that are less complex. I think we are here as startups and VCs to do the difficult things.

I don’t believe in taking the short path—doing things that are much easier. If someone is doing it and someone is super good, maybe we should buy the company. Great. Or maybe we should just partner with them. Fantastic, right?

Let’s do the things that no one is doing. That is what makes me passionate. And that is what a good CRO, I think, needs to be doing, because we’re not here to do easy stuff.

I think customer support is uninvestable today. I get so much on Twitter for this because everyone shits on them. But I'm like, Sierra and Dacorn (likely Decagon) are obviously 2 market leaders in terms of brand and how much money they've raised. You've got 16 providers who've raised over $75 million in the last 18 months, and then you've got all the incumbents—your Salesforce and Atlassian, and then your Intercoms and Zendesk and everyone in between.

OpenAI will do a customer support product if they continue doing new products. I'm not quite sure what that product roadmap is. Maybe it was going to happen, maybe not anymore. Do you agree it's not investable?

I would not personally invest. But I also think that the opportunity for companies that are established and growing a lot to go into that space is good. It's big, right? Why doesn't ElevenLabs do it?

We do. We do customer support, and the majority of our customers start with customer support, and we make an absolute ton of money on that. The ROI for our customers that use us for customer support is insane. Insane.

What percentage of your revenue is customer support? Give me 20, 30, 40?

We don't share that. But I think it works really well. It's the fastest product in terms of revenue that we've ever had. It's just insane.

But also, you look at all of our competitors that you mentioned—Sierra, Decagon, and all of the other ones—we power all of them. So we actually make money off literally customer support with our Agents platform. But we also make money from our API foundational model layer across the board. So that's the good thing about ElevenLabs: it's spread so widely.

How do you think about empowering your competitors?

I think fundamentally, we've been very happy to support the entire infrastructure layer because we believe that the opportunity in the market was way bigger than anyone was expecting, right? Voice and interactions with humans and human-technology interaction—it is the fundamental piece that was missing in the entire puzzle.

Now, you know, I'm in a situation similar to, I don't know, NVIDIA, which was competing while powering everyone but also competing with everyone, right? I think that's actually fine. It's okay that the market is big enough for multiple companies to coexist and then try to target the same customers. I'm not going to deny it. I think it also means that everyone needs to be aware that that will happen.

When we launched our Agents product, I called the biggest agent platforms that were operating on our platform, and I told them, “Guys, FYI, in the next couple of months we're going to be launching our Agents product. We're going to be competing with you, right? I just wanted to make sure that you're okay with this. I wanted to make sure that you understand that we are going into this space.”

The funny part was that all of the founders that I interacted with, after telling them this, were like, “Yeah, welcome on board. That's okay, right?” It's good because if you're able to be transparent with that and say, “Hey, I'm going to be competing with you. But at the same time, I'm partnering, and in some deals you will win and in some deals I will win, and in some deals we will partner together,” that's when you end up creating a good ecosystem in parallel.

So, I think you have to be unwaveringly aggressive now on BD and sales because I think you have an 18- to 24-month period where CIOs and CSOs are like, “AI, we have to have it. We have to have a message for it.” That won't last forever. Do you agree?

100%. Our team is dedicated to that. Everyone is selling Agents left and right. We've had the best quarter ever. It's been fantastic.

But people are incentivized to sell Agents because we know we're competing with other companies. Sometimes those companies end up complaining to us, being like, “Why are you competing with me when you're also selling to me?” But the reality is, hey, if you pay me $1 and I could be charging the end customer $20, then I'm like, I mean, maybe I'm doing something wrong here as well.

I totally get that. One of my dear friends is Jason Lampkin, who built a game and used ElevenLabs for the voice. It was great—amazing. It was such a beautiful product, so amazing, so amazing. It was expensive, and then people started using it and it became even more expensive.

He's like, “This will be the year of substitution,” where you try a great product like ElevenLabs, it's amazing, and then you go, “Oh, it's expensive,” and then you substitute for the cheaper but 80% of it. Do you think that's true?

No. I think there was a lot of conversation last year, and I've evolved my mind on all of these things. Of course, it will continue to evolve, right? A lot of people were talking last year about open-source models taking over, even in the AI voice space. “You're going to get commoditized,” and so on.

For a period of time, I kind of believed it. I was like, “Yeah, there is a point in time when open-source models will take over. It will be fully commoditized.” Great. We just need to continue going deeper on the product integrations, full verticalization, distribution, all of the stuff. Great.

What I ended up realizing is that even if open-source models become extremely good, at the same quality level that we have, it is the additional hassle, the additional component, that makes it very difficult for any organization. So, if you're a bank already, you might want to play with the idea of, “What if I use open-source models?”

The problem is that those open-source models are not built for the scale that you would want, right? You would still need to maintain them. You would still need to make them operational. And ElevenLabs actually takes that away from you. ElevenLabs Agents takes that away from you by building the entire orchestration and managing it, right?

Everyone can build anything from scratch. It's just: do you have the time, and do you have the resources? And if the priority in the company changes, are you going to be able to continue adjusting and making sure that you get the right funding in there? I don't think that is true.

We are seeing it all around the world where, like, yeah, we could go with open-source models. Excellent. Do it. You try to operationalize it, and they've lost 3 months, and then they come back.

It's interesting. ElevenLabs almost feels like OpenAI, in the way that—or Anthropic—in the way that it's this foundational layer, and you can go in any strategic product direction you want from that foundational layer. Theirs is LLMs; yours is voice.

The question is, will I be commoditized, so to speak? And then there's the question of, will I be ElevenLabs-ified? You can do SDR agents with voice. How do you think about weighing that off in terms of where your product direction goes?

I think it's an ever-evolving question all the time, and it's not that simple to decide what to do. We have verticalized sales teams right now in specific markets—markets that are much bigger or markets that are much more mature. 100%, we will have people that only do BFSI, only teams that do health care.

What does it take to do vertical sales teams, and what have been your big lessons on what mistakes people make?

I'll give you an example. In India, I tried to implement a verticalized sales team too early, and essentially depressed our revenues for a single quarter. It was an absolute disaster.

What happened there, then? You divided everyone and then it didn't work? What happened?

It didn't work because, A, people were not passionate enough. B, it was a lot more—you needed a long time to actually close some of the biggest sales. And C, the team was way too small. If you do it early on, if you segment too early, then you fundamentally end up screwing things up.

What I realized is that I have this placeholder in my calendar every single week that's called pipeline construction. That pipeline construction is similar to portfolio construction, which you're familiar with. But for me, it's pipeline construction. How do I take a VC term and put it in my mindset? How do I design the perfect pipeline for any given market and segment?

That's when you start thinking: actually, you need the liquidity, but you also need the big whales. So each one of my account executives needs to have the ability to close big whales, specifically on the enterprise segment, but also have liquidity in the pipeline so that they don't lose confidence, because people were losing confidence.

You just told me you have pipeline construction. What's the next subsequent step? You break out how many sellers you have in that market?

How many sellers? Let's say we have 20.

Let's say 20 for any given market. It doesn't really matter. I mean, we have a lot less than those ones.

Sure. Let's just say 20. What do we do then?

20? Average deal size? I look into the average deal size. I look into what the number of companies in there is. What's the average deal size?

$100K?

Depends on the market.

Let's just say $100K because it's easy for our maths.

$100K. And then we're like, okay, if I need to hit at—

What's the maths we're trying to get to? I'm just trying to understand.

So, for me, the key item is: how many companies are in that market? How many very complex enterprises are in that market? And how do I create enough liquidity in that market to close deals every single week so that the team stays motivated and sees the challenge of closing deals?

Have you ever had a situation where you're like, “The ramp I need to get my sellers on to get to my quota is too high. I'm not equipped enough to hit that quota”?

100%. The government vertical is one of those where, like we said, I thought, “You know what? I like this,” and it was one of the big bets that I made this year.

Why do you do government? In the nicest way, the sales cycles are hard and it's tough to get in. You're crushing it in other areas. Why government?

Fundamentally, one, it's very sticky. Number 2, it's a benefit that we make to society. If we can deploy agents for people to have better interactions with the actual government and government services, it's an absolute no-brainer, and I don't care if I don't make enough money to justify it or if I lose money. It doesn't really matter to me.

The third is that it's a hard industry to get into. So, the moment you actually get in, you never leave that industry. Fundamentally, when you start mapping those things out, you need to make the bets. Some of those bets will pan out really well, and some of them will not pan out that well.

It's literally portfolio construction. How quickly can I do some of these sales so that I can spend time in the difficult ones, while my investors don't call me saying, “What the fuck are you doing? I'm going to fire you,” right? I think that's a fundamental idea of portfolio construction. You do it at a country level, with each one of the new countries that you actually want to launch in.

That is a completely different concept from what the majority of CROs are thinking about these days. If you tap into AI agents for go-to-market and all the pieces, then you have something quite unique for your business.

What did you do in India, then? You verticalized too early and segmented too early?

Yeah. We depressed the market. We went back to zero. We went back to horizontal. We said, “Let's do horizontal. Let's allow people to go much deeper, and then we'll start segmenting again when the timing is right.”

What ended up happening is that we would then give them a named-account list based on how I was thinking about portfolio construction, and negotiate with them based fundamentally on that. That worked really well. We literally lost a quarter, ended up changing all of that from the ground up, and then you could see people starting to close deals in the first month.

It was insane because it was like day and night. We had changed the culture in the team in the local market by getting it wrong, then saying, “Let's go back to the basics. Let's redo it from scratch, accept that we made a mistake, and move on.”

8. Does brand reduce enterprise sales cycles?

Does brand dramatically reduce the enterprise sales cycle? When you come at ElevenLabs in India, I'm sure that over there it's not actually a very big brand.

No, but versus in the UK and the US, it's a very big brand. It's much easier.

Does brand reduce the enterprise sales cycle?

Yes, 1 million percent. Whoever tells you no is just lying. That is a fact, right? The ideal scenario is that no one gets fired. Do you remember IBM? No one gets fired for buying IBM. That is the ideal scenario for any brand. You literally become the safest asset on the block.

It's a long way. I think the 2 companies that have been able to make it today—or 3 companies, if you want—are OpenAI, Anthropic, and Cursor. Full respect for them. The 3 of them, for me, are blue-chip organizations right now. Cursor has killed enterprise unbelievably well.

Yeah, they've done super well.

They've done super well, and it's fantastic. Those are the 3 brands that are blue-chip organizations right now from a procurement-team and IT-team perspective. It's fantastic. It's great.

But everyone on Twitter has one of those funny opinions where everyone on Twitter is like, “No one uses them anymore. We all use Claude Code. You're such idiots.”

Sure, but go into HSBC in Hong Kong, or wherever around the world, or Barclays Bank in Swansea. I promise you they're using Cursor in their local dev shop. It is great. It just means that brand will help you solidify your position.

But also, the bigger problem is the stickiness of these big enterprises. That's also the unwavering mission.

Is it that easy to get into the big enterprises anyway? It's not that easy. You've done it with the portfolio companies. You know how difficult it is for them.

But once you start getting into the motion, fundamentally, as a venture capitalist, I've done it, period. I have all of the companies we invest in. We basically found them, and we do all the work, really. That's exhausting, but you do it with LPs. So imagine your LPs—it's exactly the same concept. Your LPs are the big whales.

Multi-year sales cycle.

Multi-year sales cycle. Exactly. Seriously, a multi-year sales cycle. One of our biggest LPs was a 6-year sales cycle.

Yeah, I'm not surprised. Absolutely.

How do you create enough FOMO and enough brand to actually reduce those 6 years to a lot less?

Brand does that. Exactly.

What percentage of your new sellers work out?

The majority of them.

Do they?

The majority of them, yes. Our churn rate on the sales side is actually not that big. When I make an offer, I tell every single person, “ElevenLabs is going to be extremely difficult. We are a hard company to work for because we have extremely high expectations. You're going to work a huge amount of hours. I expect full commitment.”

People love it, so you end up filtering out people who don't want to have that type of life. I think that's okay.

Is it possible to retain that at scale? You get these soft-management people who are like, “Oh no, we need to filter down the messaging.” Do you think that's possible at scale?

I think it is possible, but it's possible if you try to retain a smaller team and implement the latest technologies to make everyone more efficient in general. Of course, the bigger the company, the softer you get. But look at what Marc Andreessen was saying the other day: Companies are overstaffed by 25% and even more, right?

Of course, there's a trade-off between overstaffing and understaffing. If you're understaffed and people in sales or revenue have too much pipeline, they end up closing nothing.

How many salespeople do you need to hire this year?

We're adding about another 120 people, more or less.

How many salespeople do you have now?

It's 130 people. We'll keep it under 250.

Okay, but you're doubling the sales team.

Yes, doubling the sales team. Sales is going to grow much quicker than that.

Do you worry about that? That's a lot to bring on board, doubling a sales team. What are the pitfalls that you need to avoid?

One is dilution. If you're not up front about the expectations, you end up diluting because people come with different expectations. They essentially behave in a specific way that isn't the way you want them to behave.

How fast can you correct it?

I went to a company the other day. Essentially, it was the sales leaderboard for all of the reps, and there were 7 reps. Six reps had between 5,000 and 8,000 points—just points, just saying that—and then the bottom one had 2,000. The next-best one had 5,000. It's pretty fucking clear that this person is way off.

Do you just cut it there and then, or are you like, “Give them time. It's only been a month”?

Yes and no. For us, all of the stats are publicly open to everyone. Everyone knows how much revenue we make. Everyone knows the performance of each one of the reps.

Do you have a leaderboard internally?

Yes, we do.

How do you do that to create that competition?

We post automatically. We have a bot that will automatically post into the sales channel every single week a summary of what percentage of quota attainment year to date each rep has for that specific quota, along with all of that stuff and the forecast.

On a per-rep basis?

On a per-rep basis. Anyone can see it at any time. If you're at the bottom, you're going to be able to see that you're at the bottom.

At a practical level for me, it's beyond that. That's one component, but if you have someone building strategic accounts, then I'm not too worried. If someone is building the right pipeline, I'm also not too worried. But you need to see the action. You need to see how many meetings they're doing.

If I open a calendar and someone has a full calendar that's empty, I'm like, “What did you do this week?”

Do you worry about demoralization or losing people? I mean, every day, if I'm looking at the report and it's like, “Harry, bottom,” and next week, “Harry, bottom,” then I'm like, “Oh, fuck.” You're out. I mean, it's so mean.

But something nice is that sales attracts a different profile. Sales attracts people who are much more outspoken, people who are much more extroverted, and people who actually want to do something different. By definition, you don't go into sales if you're not willing to take the harsh feedback.

Sales reps are softer today?

I think a very large portion of them are because they don't have the environment to actually push harder. Or because they've already lost the passion in that. They just want a normal job, and that's okay. Don't get me wrong. That's absolutely fantastic as well. It is not what I'm hiring for.

What's your tell for the obsessed that wins?

9. Hiring obsessed salespeople: spotting top talent in 20 minutes

For me, you can get it very quickly, right? People that have extremely high energy, that are sharp, that are thinking quickly. And there are other people that are slightly opposite, but also they're going to be very, very deep. It might take them a little more time to actually get there, and those are less obvious. But once they do, it's insanely good, right?

If people are sharp and energetic and passionate, and they come across like someone that really wants to win, you can ask them, “Hey, pitch me ElevenLabs as if I was, I don't know, a fish.” Whatever it is. It doesn't really matter, right? Then you get a sense of how quickly they're thinking, how much passion they have, how much they've researched the company, and who they're thinking about. All of that stuff you end up having in a very short conversation. I don't need to spend more than 20 minutes to know if I want to hire someone or not.

And I will tell you, the majority of times I end up putting my feedback in Ashby. We use Ashby internally. I end up putting my feedback in Ashby during the conversation. By the middle of the conversation, I have my feedback. And I'm one of those crazy guys that writes a lot while I'm talking to someone.

I hate when it gets recorded because then people end up behaving in a very specific way versus actually being themselves. So I write a lot while I'm interviewing people, and then I have my summary. In parallel, I'm writing and I'm doing my summary, and that's it. It works really well for me.

And why do you hate the recorded element?

For me, all our calls and investor calls are recorded because it's important for us that we can share them, and I can ramp up without being in the room. My personal take is interviews: people behave differently if they think that they're being recorded versus if they think that they actually have the freedom to say whatever they think is best.

In sales, we're always asking everyone to record it, and it's mandatory for everyone.

You use Gong?

We use Gong, and then it automatically fills out our Salesforce. We try to automate as much as possible. And by the way, Lovable has the best CRM that I've ever seen. Those guys have really smashed it. Shout-out to them. Lovable CRM.

Lovable ended up building their own platform.

Can you imagine if they didn't?

We should be building our own platform as well, but we're not.

Do you buy that? Do you buy the SaaS-pocalypse—that you will customize your own software entirely in the future?

For some areas, yes.

Where will you? Where will you not?

I think for core applications, the core of things, yes. So a CRM, I will have to customize it. I will have to actually build it myself. It doesn't really matter. You don't have to spend that much time. It's just a pool of data.

Have an in-house sales team?

We haven't done it. We use Salesforce. But for some other applications, why wouldn't you do it? If it fits better in your entire ecosystem of applications, if it's easy enough, then maybe you should.

Would I build an entire Google Drive or a Gmail? No, I wouldn't. It wouldn't make any sense. But for a good amount of applications, for sure. If I was a procurement guy, I would be building procurement software for myself.

So you think the SaaS-pocalypse is overexaggerated?

I think there is a very large portion that is overexaggerated, but I think there is a real fact that people will end up building their own applications. And I think that's key. I think it's—why not?

What would you like to do with ElevenLabs today that you're not able to do because of a lack of budget?

We've never actually put any constraints on the lack of budget. If you want to do something, budget is never going to be a problem. And that's been the case since the beginning.

What is the problem? If you want to do something and you're not allowed to, why would that be?

It might well be because you haven't proven with a test or experimentation that that is the right thing to do right now. So for me, it's like, you want a budget? You want what? $100,000? You want $50,000? You want $200,000 to actually prove that things work?

Don't ask me for $1 million. Don't ask me to dedicate an entire team to it. Don't ask me to dedicate 6 months of your time to that specific thing. I'm not going to do it. But if you do it on the side and then you prove to me that you can actually make it work, and I start seeing some results, we'll scale it and we'll give you all of the budget that you want.

You said that is so interesting, this idea of, “I want to see the data before I scale it.” You now do paid, and you do paid, you know, on an F1 car.

Yes.

Remind me of the team.

The Audi F1 team.

How did you weigh up the decision to do that when it's a big investment and you had no data, honestly, to suggest it would work?

It took us a long time to actually get comfortable with the idea.

How much did it cost?

We spent a good amount of money. We had a budget in mind that we wanted to spend, and we went to different teams. We ended up negotiating with them: “This is the budget that we have. This is how much we want to be spending for year 1, for year 2,” all of that stuff. “What are the options?”

All of the teams that we spoke to, some of them were like, “No, for this, no.” Or, “For this, you don't get anything. You don't even get the logo.” Some of them were like, “Oh, we'll do all of this stuff.”

Then we ended up narrowing it down to 2 teams, and we ended up negotiating with those 2 teams in parallel. Then there was 1, which was Audi F1. We were very impressed. But it took us a while to actually be comfortable with the entire concept.

Would you rather premium branding on a worse team or secondary branding on a number 1 team?

Premium branding on the worst team. 100%. You get more exposure, and you also have a lot more creative freedom. And also, you're betting.

One of the main reasons why we ended up picking Audi F1 is because, if you look at the past, the history of Audi is that every single sports championship they've entered, they've ended up winning within 5 years. And that's fantastic. They're building the car from scratch. They build their own engine. They build everything from scratch.

They're competing with all of the big ones, and I'm a big Formula 1 fan. But they're competing with all of the big ones. I actually feel that this idea of incumbents being very technologically driven, with a really clear mindset, drive, and idea to win, is who we are as a company, ElevenLabs. But it's also who Nik and the team are at Revolut. So it fits really well with your entire narrative as a company if you try to do the work.

What was the biggest paid marketing event, channel, or activity that you did that was a mistake, and what did you learn from that?

The majority of them end up producing good returns. But when I look at the ones from an enterprise perspective that have a better ROI for us, it's dinners. When we do a dinner with executives in a given city, that has the best ROI that you could have. If I go to a trade show, the majority of them don't have good ROI.

How much does a dinner cost?

You could have a dinner with 15 people, and it costs you $3,000, $4,000, or $5,000. If you want to go bigger, then suddenly you need to shut the entire place and all of that stuff, so you will have to spend more. It depends on the city. It's not the same doing it in Mexico City, Barcelona, or London. But the ROI is actually quite solid.

The benefit of the dinners is, imagine you're inviting competitors, right? So you're inviting different customers you're trying to close. Those guys are going to know that you are actually talking to these other guys, so there's a FOMO that ends up being created in place, and it just pans out really well, right?

Because they know each other—they know the usual same ICPs. They know who's doing the PM, who's doing whatever, who's doing the sales, who's being the CTO, who's the chief AI officer, all of that stuff. They know each other. That always works really well.

The worst ROI for us has always been conferences. There's no ROI in there.

What's the takeaway from that? We should be spending less on those things and should be building our own events more.

You should be building your own events more. Yes, that's why we did the ElevenLabs event.

Insane ElevenLabs event, which kind of reminded me of some kind of rock concert meets Steve Jobs kind of event. Do you need to see, ironically, a return on that?

Yes. 100%. Why wouldn't you? It's difficult to know. You end up having this concept of influenced revenue and then direct revenue, or closed revenue, right? There are a lot of different variations of all of this that you can plan. And the event that we ran in London, the ElevenLabs Summit in London, was fantastic.

We wanted to prove that a European company can actually run these big events with so many people and put on a show that people get passionate about and talk about. We also motivated people and presented some ideas. We brought some guests to talk in front of everyone else.

What is your biggest tip for me if I'm thinking about one company in particular of ours, Solve Intelligence? It's an amazing business in London, AI for patent lawyers. They should do an incredible event for patent lawyers specifically. It would be an amazing, close-knit community.

If you were to advise them—say, they should know this if they're going to do an event for their customers, having done it so well—what would you say that advice would be?

Bring the right ICPs, design the content really well, and think carefully about the content that you want.

What would be your advice on the content? When you say, “Design it well,” what does that mean?

It has to not be salesy. I think it needs to feel natural. You should do a presentation about your company, share some insights, and talk about the things that are coming next and your vision. But if you make it too salesy, people will walk away and never come back. I think that's the fundamental trade-off.

For me, these big events are about talking a little bit about ElevenLabs. Mati always ends up doing the big speech, the big keynote. But after that, for me, it's all about how we put those partners, those companies that are building on our platform or using our technology, in front of as many people as possible so that we learn about them.

I did a fireside chat with 3 amazing companies, like BCG, NTT, and Concentrix. For me, it was more about how we enhance their appearance and talk about what they're building and what they're using. It's less about ElevenLabs; it's about them. I think you need to plan the content in a way that feels less salesy. You talk about it, but you also bring partners to validate that what you're doing is actually right.

Totally get that. You said that partners—how do you think about partnerships as a channel at ElevenLabs?

Yes.

And what are the big misnomers that people have around partnerships?

There is a concept that I really like and that we've done fairly well at ElevenLabs, which is strategic partners. Strategic partners, in the bracket that we put them in, are companies that have a CVC. I know this is going to be controversial, but I actually love CVCs—corporate venture capitalists—because they help you navigate big brands. They are your champions internally, and their incentive is to actually make it work really well for the business.

So, just for people to understand, that's like Salesforce Ventures, that's—

Exactly. Deutsche Telekom, T-Capital, which—

Siemens, I think, have one.

They all end up having one. NTT DOCOMO Ventures—almost all of them have one. But the good thing is, they know their business.

They weren't sexy up until a year and a half ago, 2 years ago. They weren't sexy. Returns were not there, as we know, right? CVCs getting returns is really difficult. They were just trying to fight to get allocations in the best companies, but it wasn't possible.

Today, we've proven with ElevenLabs that we dedicated time to actually partner with Woven Capital from Toyota, the Deutsche Telekom guys, the Telefónica guys, and Liberty Global, which actually joined our cap table. All of those things—that is a fantastic distribution strategy.

You get them to invest a small amount, and then you distribute through them.

Exactly. We negotiate contracts with them, and depending on the situation, you have one side or the other.

Do they have to bring a certain amount of pipeline or promises to get a certain amount of allocation?

Yes, they do. We're ElevenLabs, so nothing is free. But I like it because it incentivizes the relationship.

The way I organized this concept initially was so funny. When I pitched it to Mati, I was like, “Really? People are going to give us money as a contract and also as investors?” And I was like, “Yeah, let's try.”

The reality was that you end up aligning incentives. If their incentive is only to invest but your incentive is actually to close a deal, how do you figure out something that works along the way? You end up saying, “You know what? You will invest—for every $1 million that you want to invest, you need to bring X amount of revenue in the next 12 or 24 months, whatever time frame you want.” There have to be some penalties and consequences if that doesn't happen. That is the way—

How much is it? If you don't bring it, do we give you less? Is that possible?

We buy you out. Ah, yes. I think it would be perfectly fine. Then everyone is incentivized.

The logic—and I strongly believe that is the case, and that's how I pitched it to everyone—was that we want the best. So, if you invest $1 million in ElevenLabs and you bring a contract, if you help us close a contract, then your valuation goes up. You're making money from the VC side, but your business is also getting more efficient because it's implementing ElevenLabs. It's a win-win for everyone.

I totally get that. I think the other thing that you also get is a public-market messaging story, which is, “We're closely partnered with ElevenLabs, a pioneer in the industry.” For large telecom providers, wherever you are around the world, or whatever enterprise you are, that helps a lot.

100%. But the beautiful side is that you end up building new products specifically for them because you get the insights from the industry.

We went into the telco industry with no idea about the telco industry. We've actually learned everything about the telco industry by partnering with KPN, Deutsche Telekom, Telefónica, and NTT DOCOMO—all of those really amazing brands that have been in the industry for a very long time.

We work very closely with the Woven Capital guys at Toyota. We're learning a lot about the automotive industry because they're on the cap table, so everyone is motivated. It's the same with financial services.

10. Partner ecosystems: the biggest mistakes and how to do it right

There are a number of advantages to having corporate VCs on your cap table, including going into different markets, just because you incentivize and increase your opportunities to be successful.

What are the biggest mistakes you think people make with partner ecosystems?

I think people think of them as a silver bullet. “Oh, we'll have a partner and then it's done.” You need to have people who nurture those partners incredibly well, teach them, train them, and onboard them.

That's one. What would be some others from your perspective?

It takes time. It takes a long time, and that is the reality. If you're coming into the distribution side with partners, or you're building an entire partner ecosystem, thinking that you're going to be making another 20% of revenues in 2 quarters, you're wrong. That doesn't happen that way.

You need people dedicated to it. You need incentives for them because people lose interest, and you need incentives for your own team. Initially, start using SQLs—sales-qualified leads, or SQLs as sales-qualified leads—to actually track how much volume is being generated and what the actual traction being generated is.

Then, once you start having a solid motion in there, you start tracking and moving it toward revenue. Each one of the SQLs needs to be generating a certain amount of revenue.

Other than letting them invest, how do you incentivize your partners to push you?

You can always do that, and Salesforce has done it extremely well. For me, they are the reference on this. They built a partner ecosystem where they get deals from Salesforce partners and send deals to Salesforce.

In some cases, in markets where they cannot bill the customer, they actually do send the deal. They're like, “Oh, you're my partner. You will be billing the customer, and then I will bill you independently.” That's a fantastic ecosystem to be building.

How do you think about the justification of rep time across accounts? What I mean by that specifically is that any company that's very large could spend a large amount of money.

With PLG, everyone tends to start small, and then the hope is they expand much bigger and bigger and bigger. How do you think about what ACV is large enough for reps to spend significant time trying to get it?

$1,000 a month. When you say, “I'm spending $1,000 a month as a business,” your brain completely switches because you don't want to put it on your credit card anymore. At that point, you start believing, “Shit, I'm spending a lot of money on that software. I do have high expectations.”

At that point, if you have a motion from a sales perspective that is fast enough, then it doesn't really matter.

And you're ready for a weird one. One of my favorite sayings I've heard recently is, “Fear is a mile wide and an inch deep.”

Mhm.

You need to take the step to realize it's not as scary as you think. Where have you thought something was very scary and difficult, where it actually wasn't once you did it?

I had a very clear conviction that launching India at ElevenLabs was going to be extremely difficult and that we might fail.

What I ended up realizing is that I didn't know enough about the market to realize that, if you do it well, you can make a ton of money, generate a ton of buzz, and generate a lot of ROI.

Do you have to send OG team members to every market you open?

No, but we do it centrally. We always try to build a market centrally from HQ, trying to close some deals so that it's free for me to actually deploy a single person on the ground.

Sorry, how much revenue is enough to deploy a single person there?

I'm already making revenue. I'm profitable. I'm making revenue in the market. I've proven that it's a market that's getting so much traction.

Do you only launch new markets when you've already got a lot of customers and you can see the PLG?

No, not based on the PLG. I don't even look at the PLG numbers. I don't care about those. I look at the enterprises and the companies in the market. How is my portfolio construction in that market, and how can I close some number of those companies to get to a certain level of revenue?

At the moment that I get to that certain level of revenue, for me, I've nailed a product idea that needs to grow and needs to prove that it can scale and that it can have product-market fit. At least I've proven that it can start selling in there. Then that's the moment that I can get a team on the ground.

What market are you not in today, geographically or by sector, that you would most like to be in?

We're everywhere these days.

Literally everywhere?

Everywhere. I have teams all around the world. Some of them have been announced, and some of them have not been announced, but we have teams all around the world, yes.

Where are you weakest?

I think, in general, we've been weak in some of the languages, and that's entirely my fault for believing that you could sign contracts with everyone in the world, with everyone in the world speaking English only. Then you realize that it's just not possible.

Don't get me wrong: I started and opened Japan by myself without speaking Japanese.

Japan's hard. My friend is Daniel Dines from UiPath, and he told me before how hard Japan is. It's a major market for them now and it's phenomenally successful, but it's hard to spin up.

It's really hard. I started doing it about 2 years ago, going to Japan constantly myself, meeting companies, and we started generating motion. It was great. Then I hired my first rep, Sang Won, who is doing a fantastic job. He's killing it. Then I hired the GM, Jim, and the rest of the team.

That was a really hard market. You could not sell in English to everyone in the world, but you also could not sign contracts that were always in English. If you end up having a team that has mostly had experience in the US, their belief is that you can repeat that everywhere in the world.

The reality is, you look at Latin America, and less than 5% of the population speaks English. If you want to open and launch in France, good luck if you're only going to be speaking English. Good luck if you don't offer French law in your contracts. No one is going to sign it, right?

There are some of these learnings that we've had that are really tough. We wanted to sell in France, and I was not speaking French. Then let's get someone who speaks French, right? There are some of those things that we've learned by doing. It's entirely my fault, but it's also part of the experimentation, I think.

If you think South Korea's hard—

We are there. We have an office. It's really, really, really tough, but the benefit of Korea—and that's why I was saying earlier that you need to have a thesis on why you're launching in a specific market—is that they produce a ton of content. They're very innovative when they want to be, and they're producing products and services not for the local market, but for the international market.

In reality, yes, you need to be selling in Korean, but the majority of your usage will not be in Korean. It will likely be for the international market. So how do you support and map the languages that they care about so that you can target those companies? At least that's the case for us.

Are you powering the decline of Hollywood? I don't mean that horribly, but there was a brilliant article the other day on the permanent decline of Hollywood. I think it was a Wall Street Journal article, but it essentially said that the democratization of content creation has changed human attention, hence the decline of Hollywood, which is producing less and less content.

You are, in large part, powering that. I can paint a very positive case for democratization, income equality, equalizing content, and also declining Hollywood. Is that a good thing?

I don't think that's how I would put it. I don't think Hollywood is declining in the model that they have today, where very big-budget productions require a ton of people and a lot of different things. I think they're going through a crisis, but I actually think Hollywood will come out even stronger than this.

Imagine that you get a startup and the startup tells you, "You know what? I need $100 million, or I need $50 million pre-seed, just for this idea, and then it's going to take me 5 years to get it out. I don't know who's going to buy it." What are you going to say? "Why don't you raise $1 million?" That's going to be your answer: raise $1 million, prove some points, and then start going there.

I think that's the problem with Hollywood right now. I don't think it's AI voices. I don't think it's AI video. I think those are additional solutions to make sure that you're democratizing it and reducing the cost of launching new ideas into the market, but that is going to be part of the solution.

It's actually why I find defense so hard to invest in. It takes 5 years and $200 million to $250 million before you can make your first dollar of revenue on a defense product, quoting Matt Statman, who's the president and CBO of Anduril. I think he knows his stuff.

Mhm.

That's just tough for me, as a seed venture investor, to get on board with that cost curve to dollar gain. Do you know what I mean?

Yeah. That's what needs to change fundamentally. How do we create content that takes fewer dollars to actually be created? You start very slowly, but then you ramp up once you realize that there is product-market fit in that content.

I've spoken with a lot of studios over the past 3 years. I've spent a lot of time in LA and in the industry, and my pitch to them is always, "Hey, let's try to create content that will capture people's attention across the channels that they operate in today. If it works well, let's scale it." Then you have the $50 million budget.

I think that's where AI can benefit a lot, because you can iterate a lot. You can create personalized content. You can create content that people engage with. We were talking yesterday about this idea of cost per engagement. Why don't we use exactly the same framework? How much is the cost for every single person who gets engaged in your content?

11. Should operators also be investors?

That content can essentially expand to be extremely high quality as you're proving that there's product-market fit. Fundamentally, the dynamic changes.

You mentioned a couple of times that you also invest. You're a fantastic investor, and obviously we invested a lot together. Should more operators be investors at the same time?

Yes. No, no. My thinking here is that founders want to be backed by other founders. They want to be backed by the best operators. They also want to be backed by the best investors.

If you are an operator and you're willing to deploy time, not money—time—to help a company navigate and try to be successful, at some point you will end up realizing, why not actually do it from within institutional money?

Okay. Because, on the flip side, when you take institutional money, it's a great responsibility taking on someone else's money, and you have already made a promise to your employer that you will devote your best efforts and all of your energy to making them successful and making the company successful.

If you're suddenly looking at defense companies at seed, you are not doing that job to the best of your ability. I don't think so.

I think that you can do both things in parallel, and one contributes to the other. I don't think there are any conflicts in there, fundamentally. I think—

I think you can because you've reached an echelon where you can, but I don't think so otherwise. Honestly, if I were one of your sellers and I were going home and, at bath time, reading about next-generation drone companies and the cost curve to launch drone companies, you'd be like, "Dude, your turf is France and media. You could run that instead." I would be saying that if I were you, 100%.

No, but I'm getting the best out of myself by doing that. The interesting thing is, I think you put it really well on the other podcast, which is: you are the hardest-working person in the office, right? I am the same, and if you want to do both things at the same time, you need to prove every single day that you are the hardest-working person in the office.

I think that's when it works. Fundamentally, you will work really hard for the company, but you'll also work really hard for your investors and your LPs. When I was negotiating my LPA with my LPs, we got stuck on 1 clause. You always get stuck on clauses and things like that.

What was the clause?

I can't even remember. But I remember that clause, which for me was absolutely silly, and I was pushing back. For them, it was core to them. I called my main LP, which is absolutely fantastic, Senda Capital. They're really amazing guys.

I remember calling them and saying, “Guys, we need to get over this. If I don't return the money and I don't do all of these things, we're all... It doesn't really matter how we see it, right? It just doesn't really matter. So let's move on. Let's close this thing. Let's make sure that I start deploying the capital, and then we will celebrate.”

If it doesn't work out, I'm accountable for all of this. I lost all of this money. We'll figure it out, but I don't think I will lose your money. I think I will do like a 10x. I'm happy with it, and we end up moving on.

What fund would you be happy with? What do you say to an investor? “I'd be very happy with a 10x. I'd be very happy with a 6x. I'd be very happy with...”

I think, for me, the way I thought when I was raising a fund is that I would be very happy the moment that I actually have 7 unicorns in my fund.

7?

It has to be 7. I'm not going to be happy—

I don't know. It's a number that I had in my head.

I don't watch football. It's soccer—sorry.

Okay, okay. Well, he was number 7, which is why. No, but I think that's the key. People that want to achieve something different have a very clear idea of what the actual vision is.

Seed only?

I don't care.

You don't care?

I don't care, because you do the Series A of ElevenLabs with $350k, you'll make a ton of money.

Sure. I need the pre-seed.

You know, I get it, and then there's brag money.

It's okay, I got you. We need the pre-pre-seed.

At some point, unit economics also matter, but I think it's important to stay within your lane as well, right? I get a lot of Series A companies and amazing companies. It's just not my area of expertise as a fund investor, and that's not where I'm going to have a lot of fun as an investor.

Is it not?

Not for me. I love the scrappiness. I love the crazy ideas. I love the companies and the founders that no one wants to back. I like those ones. I love things that are not obvious.

We just did a deal together, and it wasn't—I mean, for me, it was very obvious. For your team, with Kieran and you, it was pretty obvious. But I also talked to a lot of other VCs, and they were like, “No.” I was like, “How is this a no for you? I don't understand it.” Those are the things that I really like.

My favorite, though, is the one we did, and they had 10 people call and say, “By the way, super interested to speak about the round.” You know that?

No.

That happened a lot.

Oh, wait, so you were not interested a week ago, and now that you know that 20 VCs and blah, blah, blah are doing it, then you're interested? No, thank you.

Unit economics—does it matter early? You said it does.

A lot of the time with AI companies, it's inverted, and actually, if your unit economics are good, it almost means that no one's using your product. I don't think it matters in the early days. I think it's more about how you get speed.

But if you're building an entirely brand-new market from scratch, I think you have the power to decide the pricing, and that pricing will help you grow quicker or will help you a lot in terms of unit economics. So the key item is: are you building something from scratch? If so, then you have pricing power.

If you're not, and you're only looking at building something that other people already have, then you're going to get screwed, and you'll have to figure out how to go deeper to be able to charge more.

What's your single biggest piece of advice for me? Pretend I'm an operator at a company. I'm a head of sales, a head of product, a head of growth—any of the coveted people that you've bought onto a cap table—and I'm thinking about doing more investing. What would your advice to me be, having done all that you've done?

Be helpful. The money that you're investing is not worth it. I mean, it's just not worth anything for any company. If you put a $5k check or a $10k check or a $50k check, it's not going to make or break the company, and you need to be very clear about that in your mind.

What are the most common ways companies need your help? What's the number one way?

Go-to-market. Hiring people.

Is it hiring sales? Hiring?

Sales hiring. For me, I go into customer calls with my portfolio companies and absolutely love it. I do intros for them. I ask them to send me their decks and their pitch, record themselves pitching, and I give them feedback.

I love it because then I can figure out what we can potentially change. Are we in the right market? Are we not? Are we getting the right traction or not? I absolutely love it. I think that's how you can be helpful with your portfolio companies in your own way.

Just forget about the money. Do small checks. It doesn't really matter. But be helpful.

The same size check every time. Do not do the whole, “Oh, I think this one's great; I'll put $50k. This one I'm not sure about; I'll do $10k.”

I think there are different variations and methods. I've not done it in my life. Essentially, as my wealth kept increasing, I started doing bigger checks, and that's okay.

Sometimes, of course, we fool ourselves, saying, “This is the company.” And you know how the majority of the time we end up getting disappointed. But we get disappointed, and we'd do it again.

What's your biggest investing mistake?

I've made a few investments where I was convinced on the market and convinced on the founders, but I did not check how hungry they were to actually do go-to-market. I ended up finding out that, yes, they were very hungry to build product and execute, but they were not hungry to actually iterate on the go-to-market. It was like—

You test hunger on go-to-market specifically?

I think it's very difficult. For me, in the first conversation, I want to understand how they're thinking about go-to-market already.

If they tell me, “Sorry, I'm only thinking about building research and doing all of these things,” fantastic. I love your product. This is great. Your mindset, your ideas—it's just not for me. Because then you're going to be fighting with them for go-to-market.

I have a bunch of my portfolio companies that actually are like that. It's great, don't get me wrong. They're doing super well, and I love them, and I'm as helpful as I can be. But at some point, the party ends, and if the music ends and you're stuck without having a product where you're monetizing it, and you're stuck with the idea that we can always make more money later on, then you're going to get—

You can invest all of your money into 1 company that you've backed. Which company?

ElevenLabs, for sure.

Kind of obviously, it can't be. Okay, it cannot be ElevenLabs. It cannot be the red one.

I love that there's a company in Spain called Theker Robotics—T-H-E-K-E-R. They're doing robotics for the manufacturing industry, warehouses, and all that stuff. Absolutely the most genius guys that I've seen in the robotics space ever. They're amazing, amazing. I would just dump all of my money in there.

How quickly do you think you know a company's good? Was it obvious very quickly that ElevenLabs was good?

Yes. ElevenLabs—I spoke to Mati for 30 minutes, and I committed.

Because for me, one of my best from Fund I, I think, was Linear. Well, yes, I know Sequoia got in very early, and it all looked great. But it wasn't what it is today for a couple of years. Actually, it was a bit of a slow burner. Do you believe in slow burners anymore?

Yes, that's absolutely fine. Not everyone learns at the same pace. Not every market grows at the same pace. I think that's okay. Life is not perfect. Why would we want to have perfect companies in our portfolio?

No, let's do the tough things, always. The reality is, for my LPs, I always tell them, “Guys, we're going to get it right and we're going to get it wrong, and that's okay.”

If I invest in a founder and it doesn't work out, and the founder wants to do another company, I believed in them the first time—why not do it the second time? So, literally, you're trying to fall in the same place multiple times.

12. Quick-Fire Round

Yes, and I will, because I believe in people. I believe that they can do something, that they can learn from those mistakes and execute and move quicker and quicker and quicker. I think that's okay.

I'd love to do a quick fire. I say a short statement, you give me your immediate thoughts. Does that sound okay?

Okay. The reason I look kind of dazed and confused there is I was trying to remember this brilliant quote, and it's like, “What would you need to achieve to be happy in 2026?”

For me, if I get 3 unicorns and ElevenLabs crosses $1 billion in revenues, I would be very happy.

Wow, three unicorns for the funds. I'll be very happy. What's the single hardest thing for you today?

Time. It's really—

That's not an answer. You do it so much.

It is. I don't have the time, and I would love to spend time with you. I work too much.

No, no, no, no, no, it is not—

I think the problem is that if you don't have enough time for certain things, then you're sacrificing your partner, your friends, all of that stuff. That's the hardest thing for me. I choose to work the amount of hours that I work, and work for ElevenLabs, and work for my funds and for my LPs, but there's a choice. Then you're choosing to do things while putting aside all the things that are also equally important.

What's your escape?

Plants. Gardening. I absolutely love it. I talk to my plants. I have so many plants at home. I talk to them all the time. I water them. I take small scissors and chop them. It's so nice.

I really like it, and that's when I'm really stressed—when I don't understand how to fix a thing—I start spending an hour or two looking after my plants. I can spend time looking at them and not doing anything, just talking to them. That gives me back energy. Weird. It's very, very strange, but it is.

No, no, there's no such thing as strange. I think we all need to be a bit more strange. What is the thing that makes Mati so good?

What makes him so good is, I think, his ability to listen to people and then make a decision, regardless of whether he gets it right or wrong. He will acknowledge if he got something wrong and then change it immediately. I think that's what I really value, because you get CEOs who will tell you, “Most of the time, I need to get it right,” and then make some mistakes in there.

For me, it's actually the opposite. You should always be getting it right, or 99% right, in 5% of the things that are really mega-important. In the remaining 95% of things, you should be getting it mostly wrong. I think that's what Mati's trying to do, and he's very successful at that—allowing people to experiment and get it wrong, then push back or run with their ideas.

If there's a thing that he believes fundamentally needs to happen in a specific way, he will say it and leave it, and everyone needs to be aligned. He has that ability to combine both things at different moments in time, and it works really well. We've seen it at ElevenLabs.

Final one for you. What are you most excited about when you look forward to the next 24 months?

I am actually looking forward to the next wave of foundation model companies.

Huh? What the—

Yes.

What?

I believe that there is a brand-new world of opportunities that we are unlocking, or that is going to be unlocked, with a new wave.

Wait, what do you mean? The next generation of OpenAI?

The next generation of OpenAI. But I actually think OpenAI, Anthropic, Google, and ElevenLabs will end up buying all of the new foundation model companies that will be popping up and creating research, paying just a few billion here, a few billion there, and just swapping them in.

Would you rather buy OpenAI at 8:30 or Anthropic at 500?

Anthropic. I've put it online multiple times. I love OpenAI, don't get me wrong. They're amazing and pushing the boundaries, but I really like how—

They are, like, still?

I think they're still pushing it, and I think they're trying to do good things for the world. But they were spread too thin. It's the problem that even we at ElevenLabs are facing, and OpenAI is facing—sorry, Anthropic is facing. Everyone is facing it.

When you start being very successful, you want to do too many things, and then you end up getting too stretched and not doing anything correctly. How do you do less and then do it really well?

But do you not think that it's been stretched too thin for too long and the Anthropic acceleration has been too great?

Yes. They need to start from scratch. For me, the key item is: how do they start building the experimentation thinking? Yes, we're still number 1, we make all of this money, but at the same time, let's go back to the basics.

What do you use?

Anthropic. I use Claude, and I'm very happy. Claude is my best friend. It's actually true. I use it all the time.

What, for personal and everything?

Personal, funds, ElevenLabs, everything.

What's your biggest advice to someone who hasn't got it set up and needs to?

I'm probably the worst one of all of them, but Claude Skills, for instance, and all of that, it's really amazing. I see some of my team members, and they've just nailed it. I'm just an amateur. They're professionals at this level.

I would love to actually be like, “You know what? All of this—I'm going away for an entire month, and I'm going to be a pro at all of this stuff.” I would love to do it, and hopefully one day it happens.

Dude, I've so enjoyed this. Thank you so much for putting up with me. You've been fantastic, and I actually preferred the second time, if I can say that.

Really?

Yeah. Nice, good.