Adarsh Hiremath @ Mercor：硅谷增长最快的创业公司｜E1261

The round is $100 million, and the price was—it was at $2 billion, yeah?

Yeah. I think we’ll live in a world with many, many models with different use cases. We’re already seeing this with a lot of application-layer companies, where they all have specialized use cases for how they want to leverage the models.

I think being a recruiter is the highest-prestige position in any company. The recruiter is the one who controls the talent inflows and outflows of every company, and pretty much you can gather all you need to know about a company from seeing the talent inflows and outflows. The businesses that succeed in a world where software costs approach zero will be built on network effects.

Thank you so much for joining me.

Thank you for having me. I’m a really, really big fan of the pod.

1. How Debating Makes The Best Founders

That is very kind of you. I did my stalking beforehand, and everyone told me about your mastery of debating. You and your co-founders, Surya and Brendan, were debate champions. How did debate prepare you for founding a company? Let’s start there.

One thing about Brendan, Surya, and me is that we actually go quite a ways back. I first met Surya when I was 10 years old, and the reason we got along so well is because we were pretty much the only elementary schoolers who wanted to compete in high school debate.

At the time, we did Lincoln-Douglas debate, which is sort of a one-on-one debate format. Surya and I actually debated each other a couple of times, and then we ended up at the same high school, Bellarmine, which is also where I met Brendan. All 3 of us were on the debate team together. Surya and I decided to do policy, so we ended up being debate partners and competing in all these national tournaments.

Debate is a lot like founding in a lot of ways. I like to think of my debate partnership with Surya as my first startup, just because we had 50/50 equity in each other’s success. If one of us were to mess up, it would tank the odds for both of us. There’s a constant feedback loop after every debate round about whether you won or lost.

Picking the right debate partner is the most important decision you can make in policy debate, and similarly, picking the right founding team is the most important decision you can make while starting a company. There’s that parallel, and then there’s just the immense amount of ownership. We both had a stake in each other’s success.

2. Do People Treat You Differently When a Unicorn Founder

At the time, you were 18 or 19—you can correct me if I’m wrong—but you, Brendan, and Surya became interested in labor markets. How did that happen?

Brendan, Surya, and I started working together without any business ambition necessarily. We just started a dev shop together. We thought, “Let’s learn how to build software really, really quickly. Let’s go to these startups, figure out what they want built, and build it together.”

What we ended up doing was recruiting these really, really exceptional people from India to help us with our dev shop. Very quickly, we realized that the software was one thing, but we had found some really exceptional people, and it was more about the people than the software.

Then we thought, “We found these people in a completely manual way. Can we automate this?” That’s how the automated candidate side of the platform was born. Very quickly, we realized Brendan, Surya, and I couldn’t scale by doing sales manually, so we had to automate the other side of the platform, too—the company-facing platform. That’s how the marketplace was born.

The marketplace is born; we’ve automated both sides of the platform. How exciting—except you were at Harvard at the time, I think. Now you have this very vibrant and working platform. Take me to that moment and the decision between whether you drop out or stick to the traditional course.

It’s funny that you say I was at Harvard. I was definitely there physically, but I’m not sure about mentally. I was pretty much doing everything I could to avoid going to classes.

I have a pretty funny story about this. Brendan would visit me pretty frequently at Harvard, and my roommate at the time, Artemis, had this really weird sleep schedule. He would go to the engineering building and pretty much be nocturnal.

The routine we would typically follow was that Brendan would visit me and crash on Artemis’s bed, because Artemis would be in the engineering building working on problem sets. Then Artemis would come back, wake Brendan up, and Brendan and I would get to work together. Artemis would go to sleep during the day. Fast-forward to today, and Artemis has joined the Mercor team.

When you’re deciding whether you’re actually going to leave Harvard, it’s one thing to say it and another thing to do it. Can you take me to that moment?

At the time, it wasn’t obvious at all that we should drop out, and I really sympathize with my parents for not approving. We hadn’t raised our seed round, we hadn’t raised the Series A, there was no Thiel Fellowship, one side of the marketplace had a little bit of revenue, and I was telling them that I wanted to abandon my degree program.

It wasn’t an obvious decision at all. But, like most of these decisions, you just make them completely emotionally. I knew I wanted to work with my best friends.

For the students who want to start a business or already have one, how do you advise them on whether to drop out or stick to the traditional path?

Oftentimes, it’s an emotional decision. You can try to rationalize dropping out or starting a company, or try to figure out the exact set of prerequisites that you have to meet.

For me, the moment I knew that I wanted to drop out was actually back when we had an office in Palo Alto. The office had exactly 3 desks: 1 for Brendan, 1 for Surya, and 1 for me. I said, “Surya, should we drop out?” He just looked at me and said, “Dude, how hard could this be?”

It wasn’t a logical argument at all, but in that moment I was just like, “Let’s do this. Let’s drop out of school.”

Where was the business at this point, just to frame it? No seed round, a little bit of revenue, no Series A, no Thiel Fellowship—nothing?

We were just 3 friends working in a small office in Palo Alto with our amazing team in India.

Take me to the seed round. How did it go? Do you remember getting the term sheet? Just take me to that, because you were 18 or 19 at the time.

We were 19 at the time, so it was surreal. Initially, we thought we wanted to base the company in New York. I’ll take credit for making the wrong call there. I very quickly realized that it was the wrong decision, but we had moved to New York before raising the seed round.

For me, the more surreal moment wasn’t when the money hit for the seed round. It was when we changed our salaries in Gusto to $500 a month. I felt like we had made it. I was like, “Amazing. We just moved to New York, and we changed our salaries to $500 a month.”

Afterward, we closed our seed round, and when the money was wired, we were just looking at the account like, “Dude…”

I’m fascinated. How was that process? Did you pitch many venture investors? Did the round come quickly? How much did you raise? Take me through it. It’s a special moment.

We raised over $3 million, and it came very quickly. General Catalyst led the round, and we really, really enjoyed working with Max and Niko.

You are one of the fastest-scaling companies in Silicon Valley, in the US, and in startups in general. It was $50 million in ARR in November—I quoted it wrongly, and you may be able to correct me—but it’s much more now. You had 30 people at the time of the $50 million, and I’ve heard a little rumor on the grapevine that you do 9-9-6: 9 a.m. to 9 p.m., 6 days a week. Can you unpack whether that’s true, why you do it, and how it works in reality?

A lot of people ask me about the 9-9-6 thing. The only reason we floated those numbers out is because we didn’t want our team working on Sundays. I like to think of the 9-9-6 stuff as more of a side effect than an objective.

We’ve carefully selected people who care deeply about the mission, and the side effect of that is that they don’t want to wait until Monday to move the company forward. People really do it because they enjoy being in each other’s presence and enjoy what they’re working on.

Do you worry about creating a hustle culture with 9-9-6?

To some extent, this isn’t unique to Mercor. All these successful companies have had pretty intense cultures historically, and it’s just a function of a startup. You have to work harder than everyone else in an obviously sustainable way to succeed.

The one thing I’ll say about that is that momentum is very, very energizing. Everyone on the team feels energized.

Everyone I spoke to also said that you’re attracting the most ambitious, young, hungry talent. That used to go to Scale or Stripe, and now it goes to you. What do you think you’ve done to create a brand where the youngest, most ambitious talent wants to go to Mercor now?

When we select people to work at Mercor, one realization we’ve come to is that you can teach people a lot of things, whether it’s technical skills, going to market, or whatever else. But the one thing you can’t quite teach people is to care.

That’s something we index on pretty heavily in our hiring process and something we really look for.

I heard that you’ve been growing 50% month over month continuously for quite a while now. That growth is insane. How does that feel internally, and what’s the first thing or 2 to break?

The way I like to think about that level of growth is that it’s basically a perpetual stress test on the business. Things are constantly breaking, whether it’s a process or the need to hire people to fill gaps more quickly than you might ordinarily need to.

The main thing is that everyone in the company needs to keep outgrowing themselves: redefining what’s possible for them and taking on new roles.

What does no one tell you about scaling that you wish they had told you?

Scaling culture is harder than scaling software. When you’re adding people to the team very, very quickly, there’s this dynamic where the culture you create with the first 20 people is, in some ways, the strongest the culture is ever going to be.

3. How Culture Breaks When Scaling So Fast

Ensuring that the culture stays strong as the company grows, does new things, and has new people enter the company is really, really challenging, but in some ways it’s the most important part of building a legendary company.

We mentioned scale earlier. One of your investors said to me that you’re mostly doing data labeling for foundation models. Do you think that’s fair, and is that a niche market or a wedge into a much larger market in your mind?

Our insight about the market is that human data and talent assessment have actually become the same thing. I can take you back 5 years, when we thought of data labeling or human data as essentially a crowdsourcing problem.

Let’s say Waymo wants a bunch of its images labeled. You get people across the world to draw boxes around stop signs to make the model better at classifying stop signs. Fast-forward to today, and the nature of human data work has changed a lot.

Now, GPT-4o or whatever model is not good in a particular domain, so we need an expert to make the model better in that domain. Figuring out who that expert should be is 100% a talent-assessment problem and a perfect application of the platform.

With a lot of the labs we work with, we’re able to figure out who the exceptional people are in very specific domains and have those people work with the labs. The interesting thing about this is that it’s essentially a forcing function on our long-term objectives.

When you think about Mercor building this global, unified labor market, what do we need to make this happen? We need tons of smart people on the platform, and we need the ability to predict job performance and figure out what those people should be doing. That happens to be the exact set of problems a lot of the AI labs are having.

When we think about the AI labs today, I heard through the grapevine that, as you mentioned, you work with some of the top AI labs. How do Mercor experts fit into these labs? What does that partnership look like? Help me understand it.

It looks exactly the same as placing someone to work at any company. Just as Mercor might work with startups making their first hires or with companies hiring in a more traditional, full-time capacity, it’s the exact same thing for a lot of the large AI labs.

They’ll hire people through the Mercor platform to essentially help with post-training models.

When you look at satisfaction on a hire basis, is 90% of a hire successful? Is it 60%? What are the metrics you track, and what’s the 1 core metric you use for the success of the business?

Customers keep growing their relationships with us, so net retention is over 100% by a large margin. As long as they keep expanding, it means that we’re doing a good job finding the right people.

When you get hires wrong, are there commonalities in why you get them wrong?

It’s all dependent on the role and what you’re looking for. At the end of the day, there might be commonalities, but it’s very, very role-dependent. There are many examples I can think of, but different companies value different things. Depending on what they value, we can correct the talent prediction.

What role are you best at, and what role are you worst at?

It’s an interesting question because we place all kinds of talent at companies, everything from software engineers to lawyers, doctors, financial analysts, and consultants.

A huge part of the Mercor platform is not building specifically for any of these roles, but instead building technology that generalizes really, really well. One example is the AI interviewer. We’ve built it so that it can immediately process someone’s background and then administer a custom interview to a person, regardless of what role they’re trying to take on, in a completely automated way.

You can literally spin up this interview in under 10 seconds. For example, for this podcast, you must have spent a decent amount of time doing research. Imagine if you could have an agent pull in all the information on someone’s profile and put together what would be the superhuman interview or the superhuman podcast. That stuff is possible now, and it’s possible for pretty much all roles.

In terms of infrastructure, what models are you sitting on top of today?

The model landscape is changing so quickly, but we leverage a variety of models and have been particularly thrilled with the OpenAI models.

Have you always predominantly been on OpenAI?

We’ve always used OpenAI in some capacity.

If you could improve any aspect of the model, what would make the biggest improvement to the business and the product today?

A concrete example would be the AI interviewer. We’ve built the product so that whenever the models improve, the experience for applicants on our platform also improves pretty significantly.

In general, this is something that’s been on our mind. There’s this huge wave of models getting better and better, and we’re asking whether we can ride that wave to make our product better and better.

We leverage LLMs and all these models throughout our product. The whole product gets better as the models get better, and 1 specific example is the interviewer.

What do you think the next generation of models will look like, before we get to training data?

The whole market is shifting to reinforcement learning. You’re already seeing this with o1, o3, and the DeepSeek models. As a result, I think we’re going to see really, really powerful models in specific domains that can reason extremely well.

That will be really exciting and unlock a huge number of use cases across a variety of industries and domains.

Do we live in a world of many specialized models that are very fragmented, or do we live in a world of monoliths with 1 or 2 very horizontal platforms?

I think we’ll live in a world with many, many models with different use cases. We’re already seeing this with a lot of application-layer companies, where they all have specialized use cases for how they want to leverage the models.

For us, it’s hiring and beating the expert hiring manager by a large margin. For another company, it might be financial analysis in a specific domain. Across each of these use cases, I think these companies will need to make their models better for their own purposes.

How fair do you think the analogy is that the model landscape will be very much like the cloud landscape? Bluntly, there will be 3 or 4 juggernauts, and it will be very hard to switch out of them. Do you agree that it’s hard to switch out of them, or do you think that, given the transience of models, it’s actually much easier and much less defensible?

There will only be a couple of companies that are able to build these foundation models that everyone builds off of. I think OpenAI is a great example of 1 of those companies, and I think that analogy roughly holds.

I don’t anticipate there being 20 companies training foundation models that can all be leveraged in the same way someone might leverage OpenAI, for example.

In terms of the post-training data side, I’d love to hear your thoughts on how much will be human data versus synthetic data moving forward.

I think a lot of it will be human data going forward. A great example of this is evals. Evals, by definition, have to be outside of model capability. In order to see whether a model is doing well at a particular task, you need to have an eval set created by humans who are better than the model at that particular task.

Humans are going to play a huge role in that. There’s also a whole set of other use cases, whether it’s SFT, RLHF, or RL environments, in how the models of tomorrow are being trained. All of those require expert humans to essentially teach the model how to get better.

To what extent would you say that data is the bottleneck that prevents model improvement more than compute or algorithms?

Data is the bottleneck. I think that would be an accurate statement.

Why, then, do so many people tell me—including Jonathan at Groq—that synthetic data is often higher quality? It doesn’t involve the dregs of the internet, like Reddit, being included in a lot of cases. You’ll see this exponential increase in model performance due mostly to using high-quality synthetic data, not low-quality human data. Why is that wrong?

The first thing is that it’s not zero-sum. Even in a world where human data is super important for the next generation of models, it doesn’t mean that synthetic data won’t also be important. Synthetic data will certainly be part of the equation.

But in a lot of ways, the bottleneck to unlocking and unleashing the next level of intelligence will be expert humans. That brings me back to the phrase you used: “low-quality human data.” Low-quality human data certainly won’t push the models to be better at anything. High-quality human data will.

Again, that’s a talent-assessment problem. The biggest lever on data quality for creating these post-training sets, for example, is finding the right people, which is really, really hard to do.

In terms of compute and algorithms, how do you think about where we are today? Are they bottlenecks, too? We’ve mentioned that data is a bottleneck. How do you think about all 3?

All of them are pieces of the same puzzle. Going forward, I do think compute, data, and algorithms will all play a part in the equation of moving AI forward and unlocking the next level of intelligence.

But the era we’re entering requires really, really expert humans to make models better for very specific use cases.

How long will that be the case for?

For a very, very long time.

Why? I thought this was what we were getting rid of.

There’s a huge long tail of tasks that models can’t do. If we reach the point a couple of hundred years from now where models are able to do every single job and humans no longer have any work to do, society is going to look really, really different.

We’re all going to be living on a UBI, playing video games all day, or whatever it may be. But until that point comes, there’s going to be a whole set of tasks that models cannot do, whether they’re specific economically valuable tasks, like the job a consultant could do, or a specific category of engineering, or even more niche things—maybe making the model better at some specific hobby, for example.

We’re always going to need to fill in the gaps, particularly in that long tail.

I’m sorry to be like, “What?” Just help me understand. You’re teaching me.

The other thing I’ll say, Harry, is that I think people are really, really in this mindset of a unidirectional relationship between humans and AI: “I can’t do something, so I give it to the AI, and the AI takes it to completion.”

I think the more realistic breakdown is that AI, for a specific use case, might be able to get us 60%, 70%, or 80% of the way there. For that remaining 40%, 30%, or 20%, you’re going to need a human to take you all the way there.

The reality is that finding that human will become harder and more valuable to do as you get further and further up the spectrum.

If you get further and further up the spectrum—if we’re able to get to 70%, 75%, 80%, 85%, or 90%—don’t you need fewer and fewer humans because the frequency is much less as you move closer and closer to perfection?

4. The Future of Foundation Models

That’s a great question, and I think that begs the question of what labor markets look like later on. The key thing is that the market will move toward specialty and sophistication, meaning the types of work we see 50 years from now will be more specialized and often require people with a higher level of sophistication in that specific thing.

When you sell to clients, what’s the moment when they say, “Wow, we’ve got to use Mercor”?

It’s usually when the first couple of candidates start working with them. We’re able to find exceptional people at the cost of software, hundreds of times over.

When you’re in that sales cycle with them today, when do they say, “We’ve got to sign up”? Is it when they see the AI interviewer, when you show them the price, or when they meet a candidate? What’s the “wow” moment for them?

It’s usually when the first couple of candidates start working with them.

5. OpenAI vs Anthropic

How do they tend to sign up? What’s the buying process? Do they buy 1 at a time? Is it on a per-talent basis or a timeline basis? How does a deal with Mercor work?

One interesting thing about Mercor is that we don’t have a sales team. There isn’t a single person who works on sales at Mercor outside of the founders.

6. Data: Synthetic vs Human

These days, what we’re seeing is mostly customer inbound. People have heard great things about Mercor from others who have hired through Mercor, and then they reach out to us and we go from there. Right now, it’s more of a bandwidth issue than a tactical or coordinated sales motion.

What percentage of hires is done end to end by software versus having a human in the loop?

On our end, the entire process is automated. That’s everything from a candidate hearing about Mercor and going onto the Mercor platform through a job listing, to us pulling in their résumé, salary expectations, and whatever else, administering a personalized interview based on both their background and the role, and allowing them to get paid for their work.

That entire process is automated.

What does the take look like on a per-candidate basis?

It all comes back to quality. Going back to the Uber example, when I get into an Uber, there isn’t that much of a difference between the 4.8-star driver and the 4.9-star driver because the unit of work is not exponential.

With something like Mercor, there’s a huge difference between the top 0.1% and the 80th percentile. Usually, for customers, it’s not a question of price; it’s a question of quality.

If we’re able to find those 0.1% or 1% of people reliably at the cost of software and delight our customers, what we take is often a second thought.

I’m sorry—what is that take, then? Is it a standardized take or a case-by-case basis? What does that look like?

It’s on a case-by-case basis. For some customers, it can be over 30%; for some, it can be less.

When you look at candidate-completion rates, how much of that is India versus the rest of the world today? I know you specialize in finding amazing talent in India specifically.

The reason we started with India is because our parents immigrated from India, and Surya and I went to these amazing schools. We started recruiting campaigns from those schools specifically.

One of the things that got us really excited about the labor market in general and the inefficiencies associated with it was that one of the best engineers I’ve ever worked with on our team was found through a Facebook ad. I manually interviewed him, and he didn’t pass the interview.

The reason we ended up hiring him is that he sent me a really, really long message about exactly what he got wrong in the interview and how to correct it. I just felt like we had to work with him. It was sort of what prompted us to start Mercor.

But if you fast-forward to today, the number-one place that workers on the Mercor platform who have jobs through us are from is actually the United States.

Percentage-wise, is it 60% US?

It’s high up there.

And on the client side, are they all US as well?

Mostly US.

7. The Future of Programming and AI

A lot of young, exceptional people are being told today that they shouldn’t study computer science anymore because computer science is becoming so automated. Forty-one percent of code is now written by AI, and in 5 years that’ll be substantially higher. Do you agree with that advice, and how do you think about whether young people should learn programming today?

My take is that programming is actually more important today; it’s just going to happen at a different level of abstraction. One could argue that the leap from assembly to Python was maybe even a bigger leap than the leap from Python to natural language.

My answer is that the way we define programming will look very, very different. It may be a person who has average skills by today’s standards in computer science orchestrating thousands of superhuman coding agents to achieve more than we thought was even possible.

8. The Impact of AI Tools on Software Development

That skill set, which we can define as programming at a different level of abstraction—programming in English—is going to be super important.

Can I ask how the way you program has changed over the last 2 years?

I definitely use a lot of the AI tools. They’ve gotten really, really good.

What do you use, and how has it changed how you work?

A great example is Cursor. A lot of members of our team use Cursor and love it. I’m one of them.

It makes doing things that would take a lot of time so simple and elegant. A great example is testing: with a couple of prompts, you can generate a more thorough test suite than anyone could have imagined for your application.

9. Why Software Will Become Commoditised

That just wasn’t possible before. Or maybe it’s bringing the same consistency from 1 part of the codebase and refactoring it for another part of the codebase. You can basically snap your fingers in Cursor, and it’ll get done today, which is absolutely insane to think about.

I think the implication for software is that software is going to get commoditized very, very quickly as these coding agents get really, really good.

What does a world where software is commoditized look like? What does that mean?

It means that people will be able to build applications much faster than was historically possible. It also means that the businesses that succeed in a world where software costs approach zero will be built on network effects.

The companies that don’t have network effects could give away their entire codebase and still be alive. The marketplaces and companies like Meta and Airbnb that have built really, really strong network effects will be the ones that thrive.

Do you agree with people who say, “SaaS is dead,” because companies will just build their own software, or do you think differently?

10. Network Effects and Marketplaces

I think what we consider SaaS will change. The next era of SaaS will be replacing entire services, whether it’s the end-to-end process of a recruiting agency like Mercor or another service that is incredibly manual and incredibly repeatable.

You mentioned network effects. If I were to push you to identify the strongest network effect you have within Mercor today, what do you think it is?

I’d break it down into 2 categories. One is the network effect of a marketplace that you might see in a labor marketplace like Uber or a marketplace like Airbnb, where every additional company that hires through Mercor strengthens the marketplace, and every additional candidate on Mercor strengthens the marketplace because there’s a higher pool of really, really exceptional people to choose from.

The second network effect, or data flywheel, is around this job-prediction piece. We’re able to see who’s performing well in jobs and the specific reasons why they’re performing well. We can use that end-to-end data on people’s outcomes to make it really, really easy to surface the person who might be best for a given role, even if they themselves don’t know it.

How do you think about building stickiness and switching costs, and making sure that the $50 million is really sustainable?

It all starts with quality. I think a lot of the greatest products or companies of our generation have been usage-based. Stripe is a great example of this.

The reason that revenue is really, really sticky is that you’re able to create these six-star experiences for customers and candidates. I think that’s 1 thing that has resulted in our very, very quick revenue ramp.

When you think about the product today, what would you most like to change that Brendan and Surya would most not let you change?

Maybe running our entire internal hiring process for Mercor in a completely automated way. Brendan, Surya, and I wouldn’t even talk to someone when they came into the office. We’d walk into the conference room to meet them for the first time, and we’d just be like, “Wow, this person is awesome.”

We couldn’t have found this person even if we spent all day, every day trying to find them. We’re getting there, and that’s just super, super exciting for us.

How do you think about the future of remote work and remote versus in-person?

I don’t think you can do 9-9-6 and do it effectively in a remote environment. I think that motivation and intensity you feel in the same room is important.

That’s exactly why we work in person in San Francisco. Brendan, Surya, and I all get super energized by being around people. A lot of our best ideas for Mercor have come when we weren’t even in a meeting. We were just sitting around, discussing things, and then you have that “aha” moment.

I think there’s something really, really special about being in person.

What was the worst product decision you made?

At 1 point, Brendan, Surya, and I all thought that chat was the future of all UI. One iteration of the Mercor product was built entirely around a chat interface. There was pretty much no other way to hire people unless you used the Mercor chatbot, because we were so bullish on chat.

We’ve come around on that. We now mix chat with other things where applicable, or leverage LLMs in other ways. But for a while, we thought the concept of a web app would be dead and that the way you would interface with all web apps would be exclusively through chat.

11. Raising From Benchmark After a Helicopter Ride

It wouldn’t even be clicking a button to hire someone. It would be telling the chatbot to hire the person. I think that’s possible down the line, but we may have mistimed it a little bit.

In terms of funding, you have some of the best people on your cap table. You mentioned General Catalyst at the start. I’m told you’ve raised quite a few rounds in quick succession. How did you think about that, and do you agree that when the money’s on the table, you should take it?

An interesting dynamic about all of our fundraising rounds is that we didn’t intend on doing the fundraising at the time. It sort of just came to us.

Going back to the example about Benchmark, someone introduced Brendan to Victor. Brendan said we were heads-down, and then Victor convinced Brendan to have a conversation with him. The rest was history from there.

How did that process go down? Brendan meets Victor, and then you guys meet Victor and have a chat. How does that go?

Brendan had the initial chat with Victor. Afterward, Brendan was like, “Okay, I’m going to get back to work.”

Then Victor asked Brendan if he had ever been on a helicopter, and Brendan said no. Before he knew it, Brendan was on a helicopter with Peter Fenton from Benchmark. We knew very quickly that they were the firm we wanted to work with.

So Brendan comes back and says, “Hey guys, they took me on a helicopter. Let’s do it.”

We had a couple more conversations with Victor and the Benchmark team, and it was clear that they were the best. We wanted to be in business with them and work with them, and they’ve just been phenomenal.

How many months later was the next round?

The next round was about 6 months later.

You didn’t need the money at that point. Talk to me about that round. How did you think about taking the money then?

Again, we weren’t focused on fundraising. We had built a business that was doing a lot in revenue. We were paying out tens of millions, and the business was doing 8 figures in revenue. We were just like, “Okay, let’s be heads-down.”

But just as we felt with Benchmark, we wanted to be in business with Susa and Felicis and the amazing team there, which made it a no-brainer.

Do you enjoy fundraising?

Not really.

Do you have a board?

We do. It’s Brendan, Surya, and me, along with Benchmark.

Is that it?

That’s it. We don’t enjoy fundraising. The thing we really, really enjoy is moving the business forward. That’s always what founders enjoy the most, so we’ve been laser-focused on that. Sometimes it just makes sense to do a round.

You mentioned 8 figures in revenue there. When you raised that round, were you aware of how fast the revenue was scaling? Were you looking at each other and saying, “This is unbelievable”?

We definitely had that moment. At the time we raised that round, we didn’t realize how much the growth was going to accelerate. We were confident in it, but the fact that it even exceeded our expectations, wrapping up Q1 of this year, is something we’re all really, really excited about.

Tell me about this new fundraise. Was this the Felicis round?

The new fundraise was led by Felicis, with some other amazing investors, including General Catalyst, Benchmark, and others, participating as well.

How much was this round?

The round was $100 million.

$100 million, and the price was at $2 billion?

Yeah.

What a phenomenal round. Really, it’s a brilliant round in terms of dilution. You dilute 5%, get $100 million on the balance sheet, and it’s a phenomenal round for a company to do.

Thank you. We’re really, really excited to be in partnership with Susa and the Felicis team. They’re amazing.

Do you need the money? What are you going to spend $100 million on? I ask this with so much respect—I like you so much—but you guys make a lot of money, and you raised not long ago. What are you going to do with $100 million?

I think it gets really dangerous when people raise money and then think they have to spend it immediately because they raised it. Our goal isn’t to deploy $100 million tomorrow.

But, Harry, I think the thing about our business is that labor aggregation and building this unified labor market are going to take a long time. We just want to make sure we have a balance sheet that’s commensurate with that long-term goal.

12. Quick-Fire Round

Let’s do a quick fly-around. I’ll say a short statement, and you give me your immediate thoughts. Does that sound okay?

Let’s do it.

What do you believe that most people around you disbelieve?

I think being a recruiter is the highest-prestige position in any company. The recruiter is the one who controls the talent inflows and outflows of every company, and you can pretty much gather everything you need to know about a company by looking at its talent inflows and outflows.

I think the recruiting function of a company is the most underrated and undervalued part.

Does the whole “we should do more with less” efficiency push, on a per-person basis, go against Mercor and the importance of recruiters?

It goes with it. Efficiency is only possible if you find the right person, and solving that matching problem and finding the right person is really, really hard, especially with manual processes that don’t scale.

Who do you think is the best person in the world at what you do, and what have you learned from them?

I’ve had this conversation with some members of the Mercor team before. One thing we like to joke about is that company executives are a lot like athletes in many ways. There’s this drive and desire to win.

I had dreams of being a basketball player a while ago—definitely not what I do today—but 1 person who I think really embodies that winning mentality is LeBron. I like him a lot.

If executives are like athletes, how do you treat yourself as an athlete?

There’s an element of pushing yourself to win, focusing on the right things, and getting better every day. That’s something I think about: How can I be the best version of myself tomorrow, an even better version the next day, and continue that process so it compounds for 10 or 20 years?

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

Part of it is the SaaS answer I gave you earlier. Over time, it’s become pretty obvious to me that the next generation of SaaS will replace entire services end to end. I think that realization has been part of the reason we’ve built Mercor in this way.

What’s 1 thing you’re doing today that people tell you to stop?

I have to be honest with you: It’s probably my Lime ride to the office when I’m running late for morning stand-up. We start at 9 a.m.

Every day, and sometimes I’m leaving my apartment at 8:55, I’ll just take a Lime and go straight down the hills of San Francisco in the most unsafe way possible. I should probably stop that.

What do you know now that you wish you’d known when you started Mercor?

I would say just how hard it would be to build a business like this. I told you, back when we decided to start Mercor, it was a complete emotional decision. Surya just looked at me and said, “Hey, man, how hard could this be?” Brendan came in with his optimism, and we just did it.

I’m thankful for that, but I didn’t really grasp how hard building a business like this would be.

If you could have anyone on your board, who would you have?

I would have to pick Sam Altman.

You can ask Sam Altman any question. What do you ask Sam?

I would probably ask him more about what AGI looks like.

What would you want his answer to be?

I think it comes more from a place of curiosity. I know Sam would turn it back on you and go, “Why didn’t you tell me first? What do you think AGI will be?”

When we achieve AGI, or sort of think about AGI, it will certainly involve doing more economically valuable work, right? When more and more and more economically valuable work has been automated to some extent—you know, research has been automated to some extent—I would broadly put that in the bucket of AI.

It’s 2035.

Okay. Final one: where is Mercor then? Paint that picture for me of how big you are, how many people you’ve placed. Where is Mercor?

I have to work backwards a little bit, right? How many job seekers are there? Roughly, put it in a couple billion. How many jobs does each person take on? People change their roles, so maybe we factor out all the jobs Mercor creates for AI agents and roughly focus on just the jobs for people.

Create a couple dozen jobs for each person. Mercor has created 100 billion jobs and has built the unified labor marketplace, meaning that anytime a company wants to hire a person for a specific job or task, they do it through Mercor. Anytime a candidate wants to consider a company for a specific job or task, they do it through Mercor.

Mercor is able to solve the matching problem across every role and every company in a seamless way.

Would you love for Mercor to be a public company one day?

One day.

Listen, Adarsh, I’ve peppered you with questions. Thank you so much for putting up with my very wayward approach to a schedule, but you’ve been fantastic. Thank you, man.

Thank you for having me. It was really fun.