Kong 是如何诞生的：API、创业硬仗与 AI 基础设施的未来

He opens a closet, and all the bananas are falling off because one guy was only eating huge bananas. He opens another closet, and there’s another mattress with other guys sleeping in a second closet. He says, “Okay, this is real. We had no money. We took the last $600 we had to get on a United flight, and we had 90 days to make it or break it. We knew that if we couldn’t raise money, we would go back to Italy broke, and that was it.”

Every year, we do the Founders Award. The Founders Award is 2,555 stock options to the best employee of the company.

Why 2,555?

That’s 2,555 days of struggle. It’s just symbolic, but it’s a way to remember seven years of struggle. Every year is a retrospective.

I think a lot of people don’t know, or maybe don’t appreciate, how fast Kong grew when it actually happened. So there were seven years of starvation, right? It basically wasn’t working.

Augusto is the founder and CEO of Kong, which was previously Mashape. We’re covering his background, from a garage in Milan to now being the CEO of a large-scale company with, I dare say, IPO ambitions at some point.

Long way to go, long way to go, but that’s one of the steps along the way.

Ambitions. All right. So you’re doing this kind of API thing. You come to the U.S. on a tourist visa.

Tourist visa, yeah.

And the idea was to raise—

90 days.

Was the idea to raise funding?

Yeah. We had no money. We took the last $600 we had to get on a United flight through Cincinnati. We went through secondary screening. We landed—we landed through Atlanta the first time. Cincinnati was the secondary room, and then we arrived in San Francisco. We had 90 days to make it or break it. We knew that if we couldn’t raise money, we would go back to Italy broke, and that was it.

So did you raise it?

Yeah, we raised it 2 weeks before departure.

You raised it in those—

Angel rounds. We’re talking about a decade ago, right? Small checks. It was a $50,000 round with different YouTube people.

All right.

There we go. Here’s a funny story.

How did you get introduced to them?

Okay, so there are 2 interesting things that you said. Can you redo it again?

I won’t be able to redo it again.

Yeah. Number 1 is, we went to Stanford Entrepreneurship Week. At that time, Stanford was doing Stanford Entrepreneurship Week in February, and all the entrepreneurs and VCs were coming there. There was this big entrepreneurship mixer, a big party. We arrived late. At that point, everything was on paper, including all the emails and registrations. We stole the sheets with all the emails and registrations and walked home with them.

That night, until 5:00 a.m., I wrote all 400 emails: “Hey, you need to know about Mashape. We didn’t have time to catch up at the mixer, but I’m happy to give you a pitch tomorrow.”

Wow.

I think 470 didn’t reply. Thirty replied, and 10 were kind of interested. In the end, 5 or 6 came to meet us.

Then one person told Kevin Donahue, who was one of the VPs on one of YouTube’s founding teams, to come to us.

Kevin Donahue?

Kevin. He later started a company for baby books and sold it. He said, “Look, I think this is something.” So he wrote a $17,000 check on our [unclear].

Seventeen thousand?

Yeah. For us, $17,000 was life versus death. We came from $600 and a few hundred dollars. He then brought 2 other people. Actually, it was $16,000, $16,000, and $16,000. We got $50,000. The problem with $50,000 divided by 3 was $16,666. There were too many 6s. We didn’t want too many 6s in the cap table, so we rounded it up to $17,000 each, and we got a $51,000 check.

The second funny part is that we negotiated and closed the deal at Travis Kalanick’s house.

Travis Kalanick’s house. I remember there was an article about you sleeping on his couch.

Yes, and the Airbnb guys as well. What happened during the negotiation was that these guys were negotiating, and they didn’t want to—

Travis had, at that time, this house up in the Castro called the Jampad, where everybody would go. Aaron Levie would go, and Drew from Dropbox would go there on the weekends. I didn’t have a place to stay, so Travis actually gave me his place to stay for a few weeks, as long as I cooked carbonara for his better half once a week. And I did that.

Then we needed help, so he said, “Come to my kitchen. We’ll negotiate.” We sat at the table. It was myself, Travis Kalanick, and the 3 YouTube investors negotiating this convertible note at something like a 50% discount—something crazy. We were desperate.

I said, “Okay, well, I don’t want to take this deal. Screw that. Okay, okay, we’re going to leave.” I was very naïve and 20 years old. I said, “Okay, we’ll come back.”

Travis came to me, put his hands on me, and said, “If these guys leave, you’re never going to see them again, nor the money.”

Wow.

So I said, “Let me do something else. You guys stay here and figure out the counteroffer.” I went to the bathroom. Travis brought me to the bathroom and locked the door of the house so these guys couldn’t go to the bathroom. We sat there for 10 minutes, went back for another round of negotiation, and then went back to the bathroom. We did that 3 or 4 times.

In the end, we shook hands on the deal—finally, a better deal: a 42% discount convertible note instead of 50%.

And no, it was a higher cap.

And that was it. Then we shook hands, and that was it.

That’s crazy. I remember you had the top Quora post for living on $13,000 a year in San Francisco. You had raised $51,000, and now you had to make that last as long as possible. Maybe talk about the next stretch.

What happened is, obviously, we had to go back. We were illegal in 2 weeks, so we went back to Italy and came back with a B-1 visa, which meant we could be here for 6 months.

But how? We had no Social Security numbers and no credit score—nothing. You enter this American system, which is very different from the Italian system, and we had zero of everything. We couldn’t pay ourselves because we didn’t have Social Security numbers, and we didn’t have a legal visa to pay ourselves.

Did you have employees at this point?

No, it was just the 2 of us.

It was 3 of you, right?

It was 3 of us. Michele was the third co-founder back then. We did about a $1,900 monthly promissory note that the company would give us, and we had to live as 3 people on $1,000 a month.

So you were living off $1,000 a month.

Three people were living off $1,000.

Yeah, in San Francisco. How do you do that?

It was cheaper then. We were living in an Airbnb somewhere for about $100.

All in the same mattress?

Yeah.

Did you have an office?

No, we were working out of Starbucks, and we were living in Valencia. I don’t know if you know Valencia. We were buying rice, beans, tuna, and pasta. We had to find the right amount of carbs and protein at the cheapest possible price. That was the combination.

That’s how we made it. We never ate out and never bought anything. We did everything in the house. We cooked in the Airbnb kitchen: rice, beans, tuna, and pasta. In fact, we ate so much tuna pasta with tomato that Marco and I would almost throw up when we saw tuna pasta, because we were running on tuna pasta every single day.

How long did this last?

That lasted from March or April 2010 for a year and a few months.

And what were you doing during this time? I know what Marco was doing: he was writing code. What were you doing?

I was writing the blog, building the website, calling prospects to sign up, talking with everyone who signed up on the website, and trying to figure things out. I was a mix of business development, HTML, CSS, and slicing. I was also talking with investors as we started to build the seed rounds.

I was trying to do some recruiting, but that never worked because nobody wanted to work for illegal Italians who might disappear. So there were a lot of mistakes and a few good things.

All right. So take us to the seed round. Somebody decided to give you a proper round at some point.

A year later, we rearchitected. We went to Honolulu, to Ala Moana Beach, and thought about what we were doing. We were aggregating all these APIs. I think we’re finally seeing it now, but 10 or 15 years ago it wasn’t ready. We were trying to build apps on the fly through APIs with drag and drop. We were too far ahead. We were missing pieces.

But we said, “Hey, we’re wrapping a lot of APIs. Actually, the assembly-line visual still holds. We just have to pivot and build an API marketplace where all the API producers and consumers can come together.”

So we pivoted and relaunched it with TechCrunch and all that stuff.

Okay. So originally, you were doing this kind of drag-and-drop, composable—

App builder.

App-builder thing. It’s funny how many companies go through this exact journey. Then you’re like, “Okay—”

With HTML5.

Wow. And then you’re like, “Okay, no, we need to build a marketplace.”

Exactly.

Because, okay, the API economy will come. It's not just apps; it will be APIs, and we build a marketplace. We launch it. We get a little traction on the long tail, and that was already the summer of 2011.

Yeah.

After the pivot in Hawaii, we launched very fast—boom, boom, boom.

Was Hawaii on the last of the money from—

No, because San Francisco to Honolulu, I think, at that time, was like $300. So with our $51K, we were living on tuna, pasta, rice, and beans in Valencia, in the Mission, in this historic building in the middle of nowhere, and working on mattresses and things.

It's unbelievable.

At some point, I remember I woke up in the morning and thought, “I need to go to Honolulu.” We could go for a walk on the beach and think about our future—or no future—because we knew this was going nowhere. Our money was starting to drain, and we needed to pivot. That's where we had the idea: let's go for the pivot, and then come back.

I remember I was building the website. The market was going crazy. The third co-founder was building all the Java backends, and we were launching fast. We got a little bit of TechCrunch press; at that time, it was ReadWriteWeb, and the other one was Mashable. Somehow, we got covered by all of them, which now seems crazy. Then, boom, we raised, and the seed-round phase started.

From whom?

At that time, we went through a lot of iterations, but where we ended up was NEA leading the seed round and Index co-leading, too. Then we got a bunch—

So in the seed, Volpi did the—

Yeah. The first checks were from George Zachary at CRV. Back then, he didn't lead the round, but he wrote the first $100K commitment. With that, he said, “We have CRV's $100K.” At that point, CRV had a seed program, which was very unusual back then. Then we went to NEA, and he said, “We're going to lead this round.” At that point, they also started a $500K seed program.

Then I met Mike Volpi, who had just moved to San Francisco to launch Index US. It was him and Danny Rimer, somewhere in there. They also started an angel-seed program, and they wrote another big check. Then we got a long tail of investors, like Jeff Bezos and Eric Schmidt, into the funds.

Bezos and Eric Schmidt.

So again, that's a story that I don't know we can replicate. What happened is, Jeff Bezos—I knew he was special about marketplaces as a business, APIs, and developers. AWS was just starting to take off. It was this hidden secret.

Was it 2011?

Yeah. Amazon was maybe a $50 billion company then, an $80 billion company. I hired the lawyer of his family office to do the seed round. As the seed round was going along, I said, “By the way, since you are also the lawyer of the family office and Bezos Expeditions, can you introduce me to Jeff Bezos?”

That's what we did. It was a pretty cool call. He was on the road with his brother, driving around Texas, and he actually wanted to put more in and invest. Boom. That was Jeff Bezos.

What do you get from Jeff Bezos? You get the brand. We got a dinner a year, strategy, and things, but obviously, it was the big brand.

Then, second, Eric Schmidt. We were working in this coworking space, and we were the only startup that stayed the latest at night after this other startup that was doing an Expedia for cruise ships. They had just raised from NEA and Eric Schmidt. At the end, they saw us working until 3:00 or 4:00 a.m. They would leave, too, when they knew we were fundraising.

The investor asked, “Who's the hardest worker?” And they said, “This guy next to us.” So they introduced us to Innovation Endeavors, and that's how Eric Schmidt invested.

Wow.

Totally unrelated.

All right, so now this brings us to 2012. What was the total seed funding you raised at that time?

At that time, it was very big. It was $1.5 million.

It's like, wow, a huge seed round. I have to say, in 2010–11, there was a lot of optimism. It's an optimistic retrospective view to say there were 3 of you on a mattress for a year and then say you were lucky.

Illegally.

That sounds pretty unlucky.

Illegally. We were going to jail.

All right. So now you have your seed funding going.

So then we raised the seed. All right, now it's real. We had to get these visas. We went back to Italy, went to the American embassy, did the letters of recommendation, and all of that.

Actually, Sam Altman was one of the guys who wrote me a letter of recommendation when he was CEO of Loopt.

No kidding.

I sat next to him on the bus when we were going to the NEA retreat. We went to Pebble Beach for the NEA retreat, and we spent 3 hours together. I said, “By the way, I'm illegal here. Can you write me a letter of recommendation for my O-1 visa?”

No kidding, and he wrote me one. So I had this big thing about how great I am. That's great—I got a recommendation. “You've got to get this guy in the country no matter what.”

So we got these 5 letters, and I got this O-1 visa. Finally, I could come here, get a WeWork space, and start to hire. We had 7 people, and Marco also got the visa. Marco had never gone to college or anything, so it was the hardest visa to get. We figured it out with more letters of recommendation. We brought him there, stayed here, built for a year, and grew.

At that time, you needed, let's say, $1 million in revenue to raise a good Series A. For a marketplace, you needed $1 million in gross volume. We were at $50K—so a lot of traction, but not a lot of revenue. We went through a 3- or 4-month Series A raise. CRV led, and then Index—

Was it Dev?

Yeah, that—yeah, because George Zachary was more of a consumer investor, so he passed it to Dev, who was more on the enterprise side. I remember Dev told me on a Sunday at 6:00 a.m. to go for a walk to decide whether to invest or not. He knew I was a night owl, and he asked me on Sunday at 6:00 a.m. to go down to Palo Alto. I had to rent a Zipcar because there was no Uber.

We went for a few walks, and he decided to invest because he was always a big believer in APIs as an assembly line. Maybe he wasn't sure the marketplace was the right execution, but he was a big believer in the theme. I think he liked us. He came to see us sleeping on the same mattress. He wanted to really see that we weren't bullshitting about the big drama story.

It was really dramatic, actually. He came to the house, because we had moved to a house in South Park at that time, where we were sleeping and working with the 7 people. He opened a closet, and there were bananas falling out because one guy only ate huge bananas. He opened another closet, and there was another mattress with other guys sleeping in the second closet. He thought, “Okay, this is real.”

Wow. How big was your A?

$6.5 million.

For you, that must have been a lot of money.

$6.5 million. I went out saying, “Let's raise a $10 million Series A.” That was the big thing at that point. We ended up at $6.5 million, and I said, “That's it.” When we saw it, it was like, “Yeah, maybe we got a shot.”

So at some point, you decided that the market wasn't working.

The market wasn't working. We raised this, hired more people, and hired 20 or 25 people—the usual post-raise honeymoon. I remember that 6 months after, the business wasn't really doing anything. There was one board meeting where we just cooked pasta for the board members, because that was 2 years before you joined us, and there was nothing to talk about from a business perspective.

The issue was, you couldn't monetize, or there was a graduation problem. These marketplaces are very tough.

When I was staying with the Airbnb founders, I learned that marketplaces could be the biggest, most powerful businesses in the world. You don't even have to innovate once you get liquidity; you can just disrupt everything. Look at eBay. You can't kill eBay.

I thought, “There are all these APIs. You can build a marketplace, get liquidity, and it'll be the Airbnb or Uber of APIs.” But we noticed that, in our case, it was a developer API marketplace. To have a marketplace work, you need to have a long tail of low-power people. If you have concentration in a few high-power suppliers, the marketplace doesn't work. Airbnb has millions of people with low power.

That was one truth. You need to have some kind of exclusivity, like the door to access that marketplace supply has to be through the marketplace. With APIs, you could Google them and go to the website; you wouldn't go through the marketplace.

The power law for public APIs at that time was on Twitter or Stripe. There were 30 that mattered and 3,000 that didn't matter much. So it was also this long-tail power. The third thing was quality. You couldn't run a cloud marketplace and actually maintain the quality of the supply. You were always blamed for it, and there was no trust.

I think because of that, it got to $1.5 million in gross revenue, but it never became the Airbnb or Uber of APIs.

That is like 10%.

Yeah. It was also losing margin on AWS, so I could never make the economics work in that model, even if this thing would have scaled. Then we go there.

It was like, okay, this is it: we're here, we're burning, and we're running out of money. So we built this massive API engine behind the marketplace API gateway that was powering 20,000 APIs in this long tail, doing billing, rate limiting, routing, caching, authentication, authorization, logging, all of that. We built it 3 times, and the third time was the great one.

We said, “Wait a second. Every company will become an API company. Why don't we take this engine and give it to the whole world?” That was the beginning of open source. And so, boom, we open-sourced Kong.

What was your runway at the time that you open-sourced Kong?

We open-sourced it in April 2015. We had to take a bridge of $2 million to go another year.

Was it an insider bridge?

Yeah, we had to take an extension because we were out of gas.

From Dev and Mike.

Yeah, they gave us an extension on the $2 million because we were out of gas. We would have died otherwise.

Wow. So you released it—I remember when Kong was released, it was actually a really big deal. When was that?

2015.

2015. Yes. Big deal. I remember it took off.

Boom.

Yeah. So this is when you and I started talking about your raise. You were raising the—

You came at the right time.

Before, these guys made no sense. Okay, now it starts to make sense. That was even a tough raise for you, because the company had been around for 7 years as a marketplace. Kong had just come out and was taking off, but it had only been out for a couple of weeks, maybe a couple of months.

Honestly, what did it for me was that we were doing all the work, and the GitHub stars checked out. Your story was phenomenal, and your command of the business was great, but it just wasn't enough. Then, while we were doing the diligence, I kept getting these kind of serendipitous signals: somebody had used Kong and loved it. I remember somebody stayed at your Airbnb and loved Kong, and then you forwarded that to me. There was just all of this zeitgeist around Kong. As a result of that, I thought, “Oh, this is clearly a phenomenon.”

Yeah. I remember I was spamming you with emails from every prospect, customer, and user. They were saying—you were just saying—“Kong usage, Kong usage, Kong usage,” nonstop. It was blowing up the inbox.

How close were we to running out of money for it to be?

After the bridge, there were 2 weeks left.

Jeez.

Nothing is ever going to stress you again your entire life, because you've been so close to being out of business so many times. I actually remember when we went to the office—it felt dead. It felt like, basically—

No, no. We were dead, and we just didn't know about it.

You were 2 weeks from being dead. I mean, it felt—

We were like, bang—you know, it flies, but it's not supposed to fly, and we keep going.

No, I remember that very, very well.

Yeah. You were like, “This guy's been there a while. I think they're going to raise money and then they're going to give up.” It was tough. And I think, if we go back, I don't know if we were able to replicate all the sequence that happened—all the things that happened.

Then we had that great sushi with Marc Andreessen, you, and me, and we sealed the deal with Marco.

Yeah. At the end of 2016, we closed, and since then the company's just been remarkable. There was an announcement recently that we crossed $100 million, which is now actually quite a while back.

A year and a half ago. Yeah.

A year and a half ago. So—

We—

Yeah. In a year and a half, it's 10x where we were when you invested. We were at less than $1 million in ARR.

Yeah. Yeah.

Less than $1 million—not even $1 million, probably $500k. Do you remember? You had a great first year, and then I told you, “Listen, if you hit $10 million, or whatever it was, I'll buy you a car.” Remember that?

I did. Actually, we have it in the new office now. I need to send you a picture.

Well, there's a funny story about this. I think you grew like—

10x that year.

So that year we went from $2 million of ARR—

Yeah.

—to $10 million.

That was right.

The plan was $6 million or $7 million. And then I checked with compliance, and I said, “Can I buy Augusto a cheap car?” They said the maximum for a gift is whatever it is. It doesn't sound like much. So I ended up buying you the best model car I could find from Japan, I think.

Yeah. Yeah. I spent a long time actually looking for this model car. Is the 280 GT?

Yeah, that's right.

So, great. I'd love to shift toward how you think about product, how you think about markets, and then move toward AI. You've actually seen a number of shifts now, right? You've seen the cloud come; you've seen the shift to APIs. Maybe just talk a little bit about how you view this current shift with AI. Is it fundamentally different? Is it the same? Does it change how you think about yourself as a leader? What's your view at a high level?

The big thing is, I'll start with API first. Obviously, we built—we became—this API infrastructure company.

Yeah. Maybe we should just describe what Kong is right now before we actually do that.

We went with this pivot, and we open-sourced Kong. The API gateway took off, and we became an enterprise company. We rebranded as Kong, Inc., and started to build all sorts of API infrastructure to run, manage, and secure your internal or external APIs.

You have software, you have microservices, and you have 10, 100, or 1,000 APIs. We're kind of the highways that make them run, and you have rate limiting. If APIs were cars, you would have guardrails, speed bumps, speed cameras, gas stations, tolls, everything, ambulances—and we provide all the infrastructure to make sure that API connectivity runs for small companies, big companies, all of that.

In this transition you mentioned, what really drove this explosion of cloud APIs was the workload moving from on-premises to the cloud. The second transition is breaking down the monolith into microservices, which creates more and more APIs. Big data and all that create event streaming. So data in motion is becoming much higher than data at rest or data in use.

But in a way, when you were starting the company, you were drafting on a big transition, right? It was the breaking down of the monolith and the shift to the cloud—

That was the key.

Yeah.

I think before there were all these archaic legacy solutions in APIs, but they weren't built for the transition to the cloud, the breaking down of the monolith into microservices, and high-scale decentralized architecture. From your time on, we kind of invented a control-plane/data-plane separation in API management, at the API control-plane level. Nobody was doing it before.

I think a lot of people don't know, or maybe don't appreciate, how fast Kong grew when it actually happened. There were 7 years of starvation, right? It just basically wasn't working.

No, actually, I know. You don't know this, but every year we do the Founders Award.

Yeah. The Founders Award is 2,555 stock options to the best employee of the company.

Yeah.

Why? That—

—is 2,555 days of struggle, to remember every day, which is 7 years.

That's amazing.

It's 7 years.

Every year the company gives out 2,555. It's just symbolic, of course—

—to remember 7 years of struggle. Every year is a retro.

And there's this crazy diving catch, which—you know—Marco did the greatest Kong. He threw it out on the market, and it caught on. Once it did, the company grew incredibly quickly.

And we had 45 competitors when we started.

Oh, I remember that. Yeah.

Now it's probably 30.

But even then, at the time, Kong broke away pretty quickly. In recent years, you've clearly been the leader in this. That only happens, I think, if there's clearly a market need. The market also kind of did you a favor in that MuleSoft got acquired, and Apigee got acquired too, right?

Yeah. MuleSoft got acquired in 2017, and Apigee got acquired in 2016.

That's right. So, in a way, there was consolidation, and you broke out as this leader. You navigated this transition from the cloud to breaking up the monolith into microservices. You built the control plane for APIs. You're considered the independent leader in the API space. Kong is the independent leader; there's no question of that.

But now you're facing another market shift, which is this movement to AI. Do you view this as directly impactful, adjacent, or accretive? How are you, as CEO, viewing this?

Yeah, I think, again, it's not like it was any of those things, but the market came at us in AI by creating more APIs.

What that means is agents are going to consume the internet in a very different way than humans did. There aren't going to be as many websites to scroll, click, and move up and down through. Applications are going to be there, but not as relevant as before.

Agents are going to programmatically exchange labor and get tasks done through programming interfaces, whether it's classic APIs or MCP protocols—which is like Duolingo for APIs, making them speak English—or whatever the protocol is. But machines consuming the internet is going to be very different from humans consuming the internet. Humans consume the internet through a UI; machines consume the internet through programming interfaces.

That's the huge shift that we're now capturing and powering, and making sure enterprises think about AI connectivity. At the end of the day, behind the scenes, it's all APIs.

We had that board meeting yesterday. What I found actually pretty remarkable is how many more banal use cases there are around AI. We can talk about agents—and I think it's worth talking about, and I agree with you—but it also feels like there's a lot of basic stuff, like key management.

A lot of companies are spending a lot of money on these AI models, so it's great to be able to have monitoring, billing, key rotation, and all of that stuff. So you're actually seeing non-agentic traffic now?

Yeah. I think the world is a bit of a stack. You have these AI companies like OpenAI that are just building models, yada yada. Then you have these wrapper companies that are trying to solve HR issues or whatever through AI. Then you have these LLM companies, like Anthropic, with a lot of revenue through APIs.

What is missing is the infrastructure to make those AI apps talk and run, and that's what we do now with AI Gateway and a lot of products. At the core, it's the boring problem you mentioned: authentication and authorization. You can be AGI as much as you want, but in the end, an agent gets stuck if it has to authenticate.

To authenticate, you've got to get an API key. You need a human in the loop to get the API key and authenticate. But if you can provide infrastructure at the beginning, you can provision and automate key provisioning, key rotation, and authentication.

Once you're there, you can unleash your LLMs and agents to just roam free without getting stuck every time on authentication and authorization, because you're already provisioning all these keys. I think that's what we want to offer our customers.

So maybe just assume for this part of our discussion that whoever's listening is pretty familiar with AI API infrastructure. Do you think that the way API infrastructure looks in 6 months, a year, or 2 years is dramatically different because of AI? Or is it that the infrastructure that we have in place and understand today evolves to also service agents in ways that are pretty obvious?

To what extent do you think this is transformational on the infrastructure layer versus more of a transition?

I think you cannot do AI if you're not API-first. You just don't have the mouth and the ears for a model. That's how AI talks.

The way we know API infrastructure will evolve; it won't change. But API traffic and AI traffic are converging. Classic API calls, yes, but also when you move tokens, it's an API call. Intelligence, let's say, will be sold through tokens. Every earnings release is about tokens, but behind each token there are calls; the payloads move tokens.

So they're kind of converging. What we're calling or building is a unified API and AI connectivity platform that helps you navigate this transition as you're managing classic API traffic, agent traffic, MCP traffic, and LLM traffic. I think it will converge into a unified platform, and that's how intelligence will move.

It's an evolutionary step over 2 or 3 years, but we really see MCP traffic growing every quarter.

I mean, I see even really basic things. For developers, this is evolving very quickly. For example, for fun, I develop using Cursor. I used to be a professional developer in a previous life.

Even knowing what APIs existed in a company required a ton of reading through docs or some weird search. It feels like now, especially if you have something like Kong, which has all the metadata and the catalog, and then you can integrate it into something like Cursor, not only does the agent itself know about the APIs, but you can expose them to developers.

It feels like a lot of the development paradigm is shifting now because we can start surfacing these things to developers.

Yeah, I think you can think of us as an enabler that helps large companies and small companies get that into the hands of more developers, or more Cursor users, because it all starts with APIs. We can provide you that infrastructure and all of that.

As you saw from yesterday, we have a very exciting roadmap that's going in that direction. Those are all of them. I think here's the big bet.

The big bet is what happened in microservices. I grew up in Italy, and you don't really study a lot of math; you just study history. But what you learn from it is that even if it doesn't exactly repeat, it rhymes. There's always an analogy.

So what we saw with microservices is that we first built rate limiting and authentication 10 times, 100 times, 50 times into the Python framework, the JavaScript framework, the TypeScript framework, the Java framework, whatever, PHP—blah, blah, blah. At some point, it didn't make any sense. Let's abstract everything to a gateway pattern and move all this connectivity logic to the gateway, and it dispatches to the right service no matter what language.

The same thing, I think, is going to happen in LLMs. At first, you have one enterprise using one big LLM. Now they're going to use 5, 10, or 100 small, medium, and large LLMs—whatever.

Once you get to that point, you don't do token rate limiting and token authentication in each LLM using the framework or LangChain, whatever. Eventually, the same thing happens as in microservices: you abstract that to a gateway pattern once again.

That's where I think AI Gateway will have the same analogy: dispatching the right connectivity logic to the right LLM versus rewriting it in each one. That's the big bet that we make—that what happens in microservices happens in how enterprises will run and govern hundreds of LLMs.

Great. Well, listen, as we wrap this up, I think it'd be great to end with any sort of recommendations for people listening to learn a bit more about how AI is shifting the API landscape, or maybe some advice on—

Don't start a company.

Maybe advice on how to start thinking about getting ready for that.

I think, as I said before, there's a lot of focus on model pretraining, tuning, and all of that, but a key thing is the connectivity layer: how LLMs, agents, whatever, will talk to each other and how you run them.

I think a lot of the traffic will be very strategic, and you have to build the stack of the enterprise to manage that AI connectivity for your next apps, your next internal tools, and your next customers.

And for the budding founders listening to the most hardcore Silicon Valley story ever, what advice do you give to them?

I think what I learned over a decade of—

Do you ever think that maybe you should have cut your losses and tried to start it over again? Do you ever think that would have been the right thing?

Never.

No.

Never. Never. The reason is I could never visualize myself going back to the airport in Italy and having my dad pick me up and say, “How’s it going?” and just failing with my tail between my legs. I could never—I would have died here without food. I couldn’t do that, and that never crossed my mind.

Shivers. That was so good.

But never—that’s the visual. Every time I even start to think about it, I have that visual and immediately say, “I’m not going back like that.”

The thing to advise is that it always takes longer than we think. It’s good to take a trend that lasts 10 or 20 years, because you have time to grow into it and make a lot of mistakes and build.

You just have to generally believe in this trend, versus falling for glitter, because it’s going to take longer than you think. As a leader, as a human, as a market, as a product, as revenue, it’s always going to take longer.

So take something that lasts and put 110% into it every day, and then it will compound. Keep the burn rate low in the early days.

Don’t die.

Don’t die. Don’t quit. Those are the things I learned along the way.

Augusto, it’s been a privilege and an honor working with you these past few years, and thanks so much for coming on the podcast.

Likewise. Thank you, Martin.