Agentic Payments: Breeze went from Zero to $4 Billion in Payment Volume in 4 Years

So we crossed 4,000,000,000 annualized volume within four years. And then three months after we started the company, FTX happened. And so we basically went back to our drawing board, talked to all the businesses we were interacting with. Payments is built on very legacy infrastructure, so you can just go and build a lot of automation. We fully indemnify the entire transaction in terms of chargeback. We fight them automatically using all these AI agents we created.

We just have twenty four seven agents responding to all these long tail edge case questions instead of businesses having to staff a team up to do it. It's like 95% plus approval rates on first tries. And we literally go in with that to, like, massive enterprises with massive teams. The checkout page was built for a human consumer. It's not built for machines. But because of all these anti bot, anti agent things that we've built in the past, the agent isn't able to transact.

My take is actually agentic payments is already here, which is apparently not everyone's take. The biggest bottleneck to, like, growth is your road map itself, and you can keep disrupting that if you can implement as much AI as possible.

It's Drew. Quick thank you to Altitude making this episode possible. Altitude is rebuilding business banking for teams on the frontier. They offer the full stack for the modern CFO and founder, a beautifully designed platform for payouts, bill pay, corporate cards, enterprise controls, and so much more. Move money with speed, yield, and total control all from your stablecoin balance. Operate at Altitude. If your business is on the frontier, your finances should be too. You can learn more at altitude.xyz.

Millie, thank you so much for being on Stable Minded. As a founder, you have so many things to be thinking about, so I really appreciate your just the window of time that you'll spend with us. When I think about the current moment of where we're at as an industry, there's, like, this crazy collision happening with, like, crypto and blockchain and, like, all of the stuff that we've been building for in recent years, meeting the commercial potential of AI and, like, all this stuff.

And I think about teams that are right in the middle of that intersection, and I think about Breeze. And I've been really grateful for our relationship and partnership so far. So excited to get into your story, your background a little bit, what you're building. But let's, yeah, introduce yourself, and we'll start with a quick intro on your name and your role. My name is Millie. I'm the CEO and cofounder of breeze.com.

The idea of Breeze is to really make payments a breeze. A little bit of a cheesy slogan. I think what we're doing is AI payments operations, and I'll get into that in a little bit. Sure. But I think like the reason why I'm even in payments is because I've kind of experienced payment issues all my life. I grew up in Hong Kong and London. I moved to The US for college. I went to Columbia and then I moved to San Francisco.

I started at Stripe as an engineer, and then with Stripe moved to Singapore, and then back to San Francisco, and then now I live in New York. Yeah. A lot of moving bounces around there. Yeah. A lot of different cities around the world, but my entire life has been filled with payments problems. Like, when I was in school, my parents couldn't pay my school bills on time, and the school would ask me about it. Right?

I would go on trips with my friends who were international on ski trips, and we couldn't, like, pay bills. Or even, like, I'm sending money from my own US account to my Singapore account, and then the bank would call me and freeze my funds. And I'm like, I didn't do anything wrong. It's myself saving myself money. Right? It was a very natural deviation of my life to, like, join Stripe as a company because I was very attracted to notion of global payments.

And then I think starting my own company in 2022 was really because I thought Stablecoins was having a moment. And it was primarily driven by, like, NFTs, DeFi in that era, and then I think it's really kind of evolved over time because I think Stablecoin as rails were very attractive to me then, and that's why we started the company in the first place. Yeah. Bring us back to Stripe. I feel like so many formative moments within that chapter. What was it like?

And maybe the compare contrast of engineer at Stripe versus founder,

What were some of the problems that you were tackling at Stripe? What were some of the things that you were working on there? And then, yeah, some of these, like, moments that you found that led you to do your own thing. Yeah. So I was at Stripe for four years. I think two years into the journey, essentially,

we were trying to launch everywhere. It was like a 100 countries. And that's actually the reason why I moved to Singapore to help with APAC expansion. And that's also where my cofounder but side story. But essentially, we were in a two people team launching Alipay to like all of Stripe. Right? But a lot of the issues we faced had nothing to do with tech and had everything to do with, like, regulatory, legal compliance problems. Yeah.

We would try to launch in Indonesia and then there would be a company that had a web site. It it's called Sendit. It had a website that was like, we are the Stripe of Indonesia. And so the Indonesian government is like, we don't want Stripe. I mean, they didn't say that, but I I felt like they basically meant it because this company creates so many jobs, opportunities within the market. Right?

And so we were constantly fighting these impossible battles to try and get banking infrastructure set up, and then we just couldn't. And simultaneously, it was the era of people would go on OpenSea and buy an NFT, click a button, and then transact. Right? Like, an Indonesian person could easily do that. You just needed stablecoins. So it felt like I was suffering with problems at work, and it felt like, hey. Blockchain solves a ton of these problems, and this is the future. Yeah.

And your cofounder was hip to hip with you, it sounds like, in that moment as well. How did you meet him? So we were in a so my cofounder's name is Peng. We were in a two people teams together launching Alipay. I met him when I first moved to Singapore. It was also COVID. So I literally moved in the middle of COVID. He's one of the few people I knew in Singapore. And then I was just, like, toying with crypto on the side just for, like, fun. And I've been in crypto since the 2017 era.

And then naturally so, I talked to my coworker about these things I'm doing because I just didn't really have any friends in Singapore. Two years later, when we started the company in 2022, it was more like I had to convince him to be like, hey, this is a real opportunity in the world. We can do this great thing together. But he's married with kids. So there was a whole, like, other battle that we had to, like, go through. Yeah. What what do you think

ultimately convinced him to go for it with you? I think he really saw the future

like, he really saw the pains we were experiencing at Stripe and he really saw the blockchain as, like, a solution Mhmm. To that problem. And both of us are, like, engineers, so everything we think about is programmatic money. How do we do things programmatically? I think even today, it's, like, how do we AIFI, like, everything we do? Right? So we're, like, very engineering minded such that anything that's like, here's a bottleneck, we're just like, can we automate it? Yeah.

So you start the company moving into Breeze a little bit. What role did Stablecoins initially play? It seems like Stablecoins, like, the huge I think we're approaching episode 60 of this podcast, and, like, the whole thread line of this has been about companies, builders, people who have integrated stablecoins into their businesses in really different thoughtful ways. We've talked with issuers. We've talked with infrastructure.

We've talked with, like, exchanges, but stablecoins have been this really unique thread through it all. Where did stablecoins first hit your radar as, like, a completely new primitive for payments? Like, what what did that look like in the beginning? Yeah. So we originally started as a crypto payments company Uh-huh. In a very different kind of era when, like,

people were saying that the future of gaming will be on the block chain. We were in Korea, and Korean studios, like, pioneered mobile games. Right? Like, free to play mobile game. And we were there and everyone was like, we're doing this. And then there would be companies that have raised hundreds of millions of dollars. Hundreds of money was going into gaming. And then it was like, okay. This this is the future. And, like, we can build infrastructure for this. Right? And it didn't have to be payments.

It was wallets. It was payment. I was like, look. There's infrastructure here to be built. Mhmm. And there are businesses here to service. And then, like, three months after we started the company, FTX happened. And then a lot of these grand plans that people had just, like, slowly evaporation into thin air. Mhmm. Right? And then we were faced with that reality of, like, okay. What do we do in this new era where stablecoins may not be front end of how, like, people consume things, but maybe it's in the back end.

And so we basically went back to drawing board, talked to, like, all the businesses we were, like, interacting with, and we're like, why do you use stablecoins? Why is it interesting? Like, how can we help? Back then, MoonPay was, like, everywhere and people were like, I kinda want MoonPay, but I I want a different version of this. And I think we just, like, kept asking them questions of what they wanted.

And then I'm gonna go to I I've heard MoonPay is also, at that point, I don't know if this is still the case, but

very expensive as well. I don't think they were known for being affordable at that point. Yeah. I think they built like a very specific niche product, which is crypto on ramp for like end users. But some people just look at this product and they're like, oh, I would love to not handle chargebacks or payment processor that settles me in stablecoins. So we then got down to the bottom of what people actually wanted.

And the way Breeze is built today is we are what's called a merchant record where the middle leg of things are in stablecoins, but, like, people are doing fiat in and fiat out. So, like, people get paid with a card and we, like, fully indemnify the entire transaction in terms of chargebacks. People could file chargebacks. We fight them automatically using all these AI agents we've created.

At the same time, they don't have to fund the payout treasury because all of that is just done automatically through stablecoins. Yeah. But consumers don't have to know that they're interacting with stablecoins. It's just Fiat in and Fiat out. Yeah. I love how you went to the customers or the early adopters at that point who,

it's always great to have people like that, like, your corner who, like, just tell me what I need to know. Like, tell me what you care about. Tell me what you value. When you think back to the what is now what seems like a very refined and clear go to market positioning, what were some of these early glimpses that you had with it sounds like an early cohort of merchants and customers where you were like, oh, this is definitely really special, or this is really the direction that we need to follow.

Were there any, like, stories or examples of conversations or partners that you felt like there was that little, like, sprinkle of magic dust

on them that gave you, this is the direction we need to go now? Quite a few moments to highlight. So because Moon Bay was so big then, all these, like, crypto native businesses, gaming platforms that we talked to were immediately like, look. I want this, but I want way better approval rates. Right? For example. Or they're like, I want that. But I'm like, why do you want that? And they're like, I actually just want stablecoin settlement. People don't really understand payments.

Like, I think if you've worked at a payments company and you've spent a lot of time understanding payments, you understand the ins and outs of various things, like how to comply with the schemes, with the banks, and the likes. What what do people miss? What's the most underappreciated

part of the stack, typically?

I think that there's a lot of, like, payments context that people miss. Mhmm. When we optimize approval rates, for example, it comes down to you have to analyze the particular bin, like, particular bank that the card is issued with. And sometimes, you can have let's say the name is passed through, it was like Milli Yang, but Milli Yang is in all caps, and suddenly they would decline the transaction. And you could retry that with just lowercase Milli Yang and the transaction would go through. Mhmm.

It's like there are nuanced things like that where payments is built on very legacy infrastructure.

Yeah. So you can just go and build a lot of automation there. Yeah. I booked my flight here with my not my legal name. Drew Andrew is my legal name. I booked it with Drew, and it was like, am I gonna get stopped at TSA? Even though, like, you know, I'm obviously the same person. Sometimes they let you in. Sometimes they don't. I can't even imagine all of the payments related exceptions or nuances that are happening behind the scenes. Yeah. What are some of these systems that you've built to increase approval rates?

Sounds like a really big value add in a way that's safe and effective.

I think it's, like, 95% plus approval rates on first tries, and we literally go in with that to, like, massive enterprises with massive teams. What's typical? People are gonna have, like, 80 something percent. And be happy with that. Like, that's kind of a Massive enterprises are probably, like, 82%,

83%. And I'm like, look. You come in with 95. It's Yeah.

We try to optimize. It depends a little bit on traffic too. So some people run a bunch of meta ads and you do get bot traffic. Mhmm. So it's like legitimate traffic. But in terms of how we've optimized approval rates, it's really like you look at a single transaction and you could retry the transaction. It could be something as simple as retrying the name in different ways you write it or it could be 3DS is not available somehow with this particular bank because banking infrastructure is just kind of old.

They're not responding consistently. But there are, like, various retry logic that we've basically built, leveraging AI, a lot of it. And so that's why we're confident with that thesis. Yeah. And that's all happening, like, within split seconds. Like Yes. Extremely quick programmatic if then retries, like, that's all happening Yes. And I think similarly the opposite of that. So some people are concerned that if my approval rates are high, then my fraud is gonna be high. Yeah. Right?

Because you're letting in traffic that you may want it to reject. So we basically had to go figure out chargebacks and reason codes, what are we comfortable underwriting in a single transaction versus not. When we sell our approval rate product, we sell with chargeback indemnification because it's a little bit of a you wanna let in enough traffic, but not like bad traffic. What does chargeback indemnification mean?

How would you Like, we literally absorb the chargeback losses of transaction because we're very confident we'd win it. I was talking to a merchant in Asia about a year ago. This was in retail marketplace.

Yeah. It was like a high end shoe company selling, like, really nice shoes, and they said that their chargeback rates were over 50%. That that's insane. Absolutely insane. 50% or more people buy their products and then, like, intentionally will either return or charge back them. It wasn't just returns. It was, like, very, very painful. It's like I can't even imagine that level of payment operation headache.

What other parts of payment operations do you find are the most, like, for the operators or merchants or businesses that you're working with? What are some of these other things that are sticky, hairy,

painful things that you come across? It could be like, if you think about you transacting on any website, when your card gets declined, you want to go talk to someone. Right? You're like, why is my card getting declined? What do I do? And you're just like, I'm trying to pay for this flight. Right? We have done a lot of work on our side to provide real time actionable feedback.

If for a given merchant somehow they have specific configurations that they can't accept certain card or the issuer is just declining transactions then, then we're like, go and retry with another card. Something as simple as that, real time feedback, actually really help increase that approval rate as well. But that's also in the realm of customer support. Traditionally, so companies have had to build massive operational teams to be answering customer decline questions.

Or it could be like in payouts when they're like, I wanna change my bank account, and then there's a million steps you have to hop through to get that done. Mhmm. So we also programmatically have built a lot of flows there. What are you most proud of when you think back to these key pieces of the puzzle that you've built? What are some features or or milestones that you think about that feel really powerful or unique? So we cross 4,000,000,000 annualized volume Woah.

Within four years, which is like something I That's a b. That's a b. Yeah. Yeah. Got a b. The b's on the table now. Yeah. Yeah. It's it's kind of insane to think about because it's a direct measurable metric of we've gotten this far as a company. Once upon a time when we started the company, three months later, FTX imploded. A few months after that, we were back to the drawing board. Right? Like, we had nothing. And at some point, we just figured it out, and then we're just able to constantly repeat that motion.

We recently partnered with Unity, be, like, their official verified solution partner to help a lot of games, external in app purchases, alternative. But it's, like, things like that that is crazy to think about. We're now a way more legit company. We've partnered with a public company. There are many moments to celebrate. But at the same time, I think when you're building a company, you always feel like you can do better. So I'm always like, can we do more? So, yeah, it's kind of always like day one. Yeah. Okay.

So you started this business. There were some really key, like, back to the drawing board moments. You've also built through what has been such a wild period of AI and the role that AI plays in money and programmatic money. What has that been like? As a founder, you've been building this in real time. You've seen artificial intelligence explode in real time as you've been building. What has that been like, and how are you thinking through applying the upsides of that technology to Breeze?

There's a lot to unpack there. There is a component of how we've adopted AI internally in the company that I could speak about. There's also agentic payments, how we're approaching that, and where I think the future is. I can start with agentic payments, seems like the hot topic right now. My take is actually agentic payments is already here, which is apparently not everyone's take, and I think apparently a hot take. But I think it depends on how you think about payments. Yeah. How should we be thinking about it?

Many people are like, when ChatGPT launched, they were like, hey. I wanna put our credit card down and then ChatGPT and go buy a $20 T shirt for me. And then at some point last year, I think ChatGPT killed this product product because I think Walmart was like, hey. It's three x worse conversion. We want to actually own the consumer relationship. This is not happening. Right? And then some people have said, oh, AgenTic payments is dead or is not there or whatever it is.

But the way I think about it is AgenTic payments has actually already been here for a while if you look at trading platforms. So if you look at Interactive Brokers, Robinhood, Polymarket, Cauchy, prediction markets, what happens is institutions or consumers basically fund a wallet of some sort with fiat or crypto. It doesn't matter. And then they call a bunch of APIs to trade. Right? And a lot of these APIs are driven by algorithms. Right? It it could be like, oh, Jude Bellingham just scored a goal.

It looks like he's winning this match, so let me go trade on CaoXi for this. Different signals are Yeah. Yeah. Signals that they're ingesting. It could be, oh, there's Wall Street Journal headline on a war is about to break out or something, and then stock traders would go sell like S and P or something. Right? But if you think about that, that is a robot trading. Right? And that's a robot transacting with money buying and selling. It's just that the constraint here is not that it's all open.

It's very constrained within a platform. So it's constrained within a predictions market platform or it's constrained within a trading platform. But I do think it is the closest analogy to agentic payments. Now I think the next question is what that looks like in the future. Right? So I think it's actually really interesting because I have really thought about this, and I think what is really interesting to me is we do a lot of anti bot things on the checkout page. Anti bot things. Anti bot things.

Anti Asian, anti bot things. How do you define what does that mean? So, like, if you go on a checkout page and you're coming in from, a foreign country or, like, a random IP or whatever, you would see, like, a Cloudflare CAPTCHA. Yeah. It's like, hey. Verify that you're a human. That's anti agent type of flow. And then you have three d s, which is, like, almost a two FA on payments itself when they think, hey.

You're transacting with a larger dollar transaction, and then they have to verify that you're a human so you get an SMS to verify, this is me. Those things are all anti bot because the assumption was that bots were malicious, but we're entering into this era when we think of agents as all good. Right? Like, I'm trying to transact, and it's a legitimate transaction. But because of all these anti bot, anti agent things that we've built in the past, the agent isn't able to transact. Those are barriers. Yeah. Naturally.

So, like, that checkout page has to evolve. At the same time, you don't wanna accept fraud. So there's great tension there in the traditional rails. How do you think about the way that

checkout is gonna change? Because I feel like we've lived in this twenty year basket of goods that you buy Yeah. On a website. I feel like that's gonna change pretty significantly in terms of how we actually purchase and transact with things. Because an agent isn't gonna go through that flow, I would imagine, or need to. And maybe speak a little bit since you brought up Cloudflare, like some of these protocols or approaches to payments for agents that are more native. How do you see that?

So I think one very interesting thing that's happening, and two, I'll answer the checkout question,

that the unit of commerce is just changing. Used to be that, like, we'd buy a $20 t shirt or, a $1 in game digital good right on the App Store. But now the way machines charge is 0.0001. Mhmm. Right? It's, not even a bip, not even, like, a hundredth of a bip, could be a millionth of a bip. Right? And fiat rails were just not built for that. Like, fiat rails were built with some sort of flat fee, some sort of exchange rate, some sort of variable fee. People would make negative margins doing those transactions.

Would make sense. Yeah. Even if we're talking about, like, picks in Brazil, like, or whatever it is, we actually literally need to get to the bottom of the cheapest rail that is available here with $24.07 settlement,

which, in my opinion, is the blockchain. I feel like that's why gaming was so hot in that period because gaming is such a great paradigm to start thinking about value transfer with all these different micro transactions and streaming value, and it makes sense why gaming was this like, oh, that makes sense, that use case. What do you think about the maturity of that or the evolution of that into more of really where you see you mentioned trading, high frequency use cases with that.

What other use cases or practical ways are you seeing agents or programmatic commerce happening practically speaking? Do you see this more as, hey. Maybe this isn't chat and Walmart for t shirts on the retail side, way more b two b or, like, what are some trends or things that you're observing right now? What is interesting is I think consumers are actually quite split on how they think about AgenTek payments. I myself love going on Claude to plan a trip,

and I'm like, take my credit card. And then Claude is like, I won't take your credit card. I'm like, why would you take my credit card? Just take my credit card. Go do this thing. Right? I live in 2026. That's how I think about it. And yet I have to go on the manual website, type all this information. Right? People do approach things very differently. Some people are very anti that, and some people could be pro that. Right? So I do think the checkout page itself has to evolve.

The checkout page was built for a human consumer. It's not built for machines. Right? If you think about how even websites today, if you go on, like, I think it was Linear or Vercel, they literally have a version of their website that's human readable that you can go on, and then they have a markdown of a website for machines to read. And we did the same thing at Breeze. We had to create a machine readable website because it helps with geo. And the machine readable website is a markdown file.

There's not many images. Right? It doesn't appeal to you. There are no animations. You're not, like, triggered to go and look at things. Yeah. Instead, it's just content. Give you a thing about the checkout page today, it's all optimized for, oh, how does a human think about clicking this button versus that button? How many clicks do they have? Machines don't care about that. Machines just wanna transact. Right?

Instead, you have to basically remove the friction that we discussed earlier, which is like the captchas, the three d s. And I do think blockchain transactions themselves is very much built for this. I can just send money and I can confirm that you've received money and this is like unlocked. Yeah.

It seems like getting this right over the next few years for the really the merchants, like companies, whether they're big or small, are building their business, they're selling things, they're what would have been marketing themselves on a website. Getting this right,

how you present yourself or be discovered by agents for commerce seems like one of the most interesting lanes to go learn more about because right now companies have spent decades positioning their entire platforms for human consumption and human readability. What are some of these things that you're seeing modern enterprises do correctly?

And maybe this is kind of where Breeze comes in a little bit with, like, how you equip them, how you work with them, stuff like that. Yeah. We don't directly touch the marketing side of things today, but I have seen that because we work with a lot of companies that have a lot of consumers. Right? And we service consumers. To us, the most important thing is the customer satisfied.

Because if the customer is satisfied, meaning they get the responses to their questions answered quickly, then they'll leave a good review. If not, they go on Reddit and they complain. And, actually, Reddit is apparently where a lot of LLMs are ingesting a lot of information. So whenever you go on, like, Plot or OpenAI, Gemini, whatever it is, and you ask them, hey. What do you think about this business? A lot of the times, they're pulling in feedback from Reddit.

And so that has a direct correlation to how good your customer support stack is. Is. So I think the only way in which we're helping businesses in that regard is we just have twenty four seven agents responding to all these long tail edge case questions instead of businesses having to staff a team up to do it. Right? And then that in turn means they get better review and better feedback, and that directly helps with their geo. Yeah. I know you had some takes on stablecoins before we move on.

What do you think on bull versus bear case for stablecoins?

You've gotten a very front row seat to the benefits, the overhyped nature of them, what they do well, what they don't do well. What is your take on stablecoins?

Are they overhyped? What's your current standing on that? The most interesting thing about stablecoins to me is that the blockchain itself is a double entry ledger. Right? And then The US financial system, if you look at it, is running on banking hours, correspondent banking infrastructure. It is a very old grail. Mhmm. Right? So it's all crazy to think, hey. The future rail looks way more advanced. We live in 2026.

That itself is probably, like, the bull case for stablecoins and the reason why genius act is happening. Everyone's, like, running to it. So I think the bull case for stablecoin is pretty obvious in my opinion at this point. I think the bear case of stablecoin has something to do with cybersecurity. Yesterday, I think last night, OpenAI had a cybersecurity bug related to Hugging Face and OpenAI, where somehow using, like, 5.6 SOL, the program was able to escape isolated sandbox.

Even in crypto, there was a Zcash minting bug quite a few months ago where they were like, oh, there's a vulnerability that was found in the contract that wasn't discovered. So I think the kind of bear case for stablecoins, in my opinion, is that as, like, AI is getting more and more advanced, agents in general are uniquely designed to spot vulnerabilities.

And the reason for that is because robots can run, like the the entry point for a program could be here, and then the exploit could be here, so it's eight steps removed. Right? And then the human, when writing tests, don't think eight steps ahead. Instead, they probably tested, like, step a b c instead of the full chain. And then suddenly, robots are uniquely great at, like, running every single test case possible to exploit your program.

And then it could be exploited, in which case, there was a minting bug with Zcash. What if there's a minting bug with, like, stablecoins? Mhmm. Which is wild to think about, but it takes a few of these headlines to hit for people to doubt the beauty of tech itself. Yeah. Yeah. And it has nothing to do with stablecoins technically. It has something to do with the the rails it's run on. It could happen. So like traditional payments infrastructure,

but it so happens that smart contracts are published like outright. Everyone can go take a look and everyone can figure out how to exploit the contract. From your vantage point, Breeze works with both fiat, stablecoins and, like, you're gonna deal with both. You're gonna have to. From your vantage point, what are some areas that both uniquely shine compared to the other? Like, what are some things that you're like, oh, yeah.

If we had a choice or if we were to recommend using one or the other for certain things, like, there obvious

examples where one wins or loses in your mind? The fiat one is actually pretty easy. I was having this conversation today with a travel company, like hospitality in general. It's very interesting, like, rails. That's, like, thinking about stablecoins. But I think one of the biggest challenge I got was, Millie, there's only USDC and URC. I have all these different currencies. How are you gonna handle those? Yeah. And I don't want them converting back to US dollars or euros and then converting back to, local currency.

It's extremely inefficient, which I think is the unique nature of fiat. We do have multiple currencies. Yeah. And the world is does not just run on US and EU rails. Right? That's like one unique benefit of fiat. Another unique benefit of fiat is that people can spend that money really easily. It's very easy to go get a card issued on your bank account and go spend it. Whereas, when you have a stablecoin wallet, it is still very difficult to go spend that money.

So the liquidity is there, but there are a few steps removed. The beauty of stablecoins, however, in my opinion, is you go pay for a transaction, you wait a few seconds, then it's done. I think it's the most beautiful thing ever. You go buy a digital good, you pay, and then it's unlocked, and you see it. And then as a business, I would love to receive the money earlier and I can earn yield on it. I could do a lot of things off it. I don't have to wait until traditional rails settle me in t plus one or t plus two.

Yeah. And then I think another unique benefit that a lot of businesses that we service use is actually that they don't have to reserve over the weekend liquidity for payouts because they're like, oh, I can just fund you in stable coins whenever and wherever twenty four seven. I don't have to go reserve three days of balance with you. Mhmm. And programmatic like dollar, I think, yield bearing stablecoins, very attractive. Yeah. You don't have to manually go send money to a yield bearing account and then send it back.

I think it's very obvious what the benefits are

Mexico City recently, and we were meeting with a couple teams that were really focused on that FX piece or local fiat piece on, like, tokenized sovereign bonds, for example, like with EtherFuse is doing some really interesting work with that. When you were saying that, I was like I was like, the Dave Taylor EtherFuse bat signal is, like, going up.

The problem that I think they're going after is, like, if we can bring the localization of money on chain in a way more efficient way, we're not having to go in and out all the time, but there are obviously some hurdles hurdles to that. What are some other, like, maybe misconceptions that you see people have, especially in the relationship to programmatic commerce and AI payments? Kind of asking this for myself. Like, where are people applying stablecoins where they're not really needed?

I think there's a lot of hype around micropayments and stuff that Cloudflare is coming out with, x four zero two, all this stuff. What if this is actually solved already with nonstablecoins

versus, like, oh, that's really interesting. Well, I think the primary use case for Stablecoins for us right now is as a substitute to a US bank account. So a lot of the foreign businesses we service want Stablecoins because then they don't have to manage a US bank account, account, which is true of the merchant of record solution. You're servicing foreign companies.

The foreign company doesn't want to establish a US entity, US directors, whatever, and they need a bank account, and then they can't get a bank account. Right? So stablecoins in general are uniquely good at that. Where it isn't great I I think on ramp and off ramp fees is probably the biggest headache. What Sablecoins are you mostly dealing with at Bruce's? USCC. USCC. Yeah. Have you considered other approaches like issuing your own or using others? Yeah. We are going to.

The reason why we haven't done that is really because that's always, like, new product priorities that come up. But, essentially, if you're looking at one to three days of flow and you multiply that number by 5%, you have to run a ton of volume for that to make significant revenue. But I do think, like, the prime example I love to give is company like Airbnb. The company actually has money transmitter licenses.

And the reason why it went and got money transmitter licenses is because when you pay for an Airbnb travel booking, you don't actually stay until three or four weeks later. So what happens during that period of time is Airbnb holds your money. That float is great. Yeah. Yeah. Yeah. And I think it was something like 225,000,000 or something like that in like three months or something. Yeah. Or is it a year? Massive bank. Yeah. They built basically I mean, Starbucks also. Right?

So, like, Starbucks is a whole, like, different program. But Airbnb's program right now for marketplaces in general, like, you can build out with stablecoins very easily, and you don't have to go get your own money transfer licenses. You can go partner and be like, hey. This is the back end currency that's working under the hood. So maybe that is actually the point where a lot of people are like, oh, I'm trying to launch my own stablecoin.

And I'm like, does the stablecoin have to be front end that consumers can see, or does it just exist in the back end? Mhmm. Airbnb could launch their own stablecoin in the background, but the consumer doesn't have to see it. It's just an internal treasury operation to, like, optimize on yields, but the consumer doesn't have to pay with Airbnb USD. I do see, like, a lot of people rushing towards, hey. I'm launching my own stablecoin. Like, we get pitches like this all the time. From the, like, infrastructure side?

Yeah. Yeah. It's like we're from people who are issuing new stablecoins, they're like, why don't you support the stablecoin on your checkout page? And I'm like, what liquidity do you have on your stablecoin? Like, who's using it? Right? And I think that they're trying to build liquidity because clearly Circle had a phenomenal IPO. Tether makes a ton of money, apparently the most per capita of like any company in the world. So everyone's looking at this opportunity and they're like, I want this opportunity.

Right? But there's no clear reason why they have to go create a new stablecoin. How do you think about the because you mentioned, like, just thinking about what it would look like for Breeze to issue their own. It doesn't really make sense to go and, like, because right now, we've seen this.

A lot of companies actually, majority of them will start with USCC or USTT because of the all these reasons, and then they'll get to a point where they're like, you know, it probably makes sense for me to, like, own a little bit of this upside and have a little bit more control, and they'll typically go to the issuance layer of the Braille's or the m zero's or the some bridges, Paxos, Anchorage. Like, when you're thinking about that decision, what's really important to you?

What matters in terms of who you might work with there? Genius, OCC,

Charter. I think everyone's rushing to actually do this right now. And Agora. Forgot about you guys. Sorry. Of course. But I think that there are companies with conditional approval. I think Anchorage is the only one that's actually like been operational with an OCC charter Yeah. They checked all the boxes. Or something. I think. Yeah. But I think that there are many people that are rushing to do that. To me, it's more like I'm trusting someone else to hold my dollars. Right? So my dollars better be secure. Mhmm.

That's how I think about it. So that's probably like first and foremost and obviously fees. But I think it's those two and and third was like infrastructure like, can I trust it with reliability? Like, it has to be always available. I have to be able to get my cash. It's a little bit like opening a new bank account Mhmm. And putting a bunch of money there. Yeah. Yeah.

Okay. So let's bring this back into the current moment with Breeze, AI payment operations, and where you go next? You're talking with, at least today, you're probably talking to a lot of different companies, but a travel marketplace or company in that world. Like, who are you mainly wanting to speak with on the commercial side of things? What is the perfect bull's eye customer size example of, like, if this is you and you're listening to this podcast right now, please fill out our lead form.

We would love to talk to you and sell you Breeze. We're actually talking to a lot of enterprise companies

that have somehow not really established a massive payments team. So when we talk to, like, massive enterprises, I can't really name them, but, like, the most like, the biggest, like, enterprises you can think of, a lot of them have massive payments teams. So, like, a 100, a 150 people. And then whenever we talk to them, they're like, okay. Approval rates, we've solved it. Chargebacks, we've solved it. Like, all of these things. And it's like really tough.

Are like so you have a 100 plus people in these big organizations. What are they typically doing? Like, what are the core roles that these teams are are They they could be like launching new markets. They could be analyzing declines. They could be doing customer support ops, chargebacks, KYC related operations, compliance, and the likes.

I think we really nail it with companies that are, like, either hyper growth, so they've managed to grow to certain scale without scaling up payments team, and then they're now thinking about it. Because they're like, look. If I can lift my approval rates by one or 2%, I can make this much more revenue. That's, like, perfect for us. Or, like, it could be enterprises that are launching in new markets, and they're like, I want to quickly test Australia, Japan, Korea. We can do all three of them. And we're like, okay.

We'll be your team there, basically. Yeah. And it's complementary to their current offering. It's just their finance team doesn't wanna go a whole JPY, KRW, like, I don't know, Indonesian rupee or whatever. A lot of these kind of either hyper growth companies that are like, they've grown without thinking about payments and then now they want to think about payments. It's it's a pretty perfect fit.

Or massive enterprises where they're like, oh, I want to launch in a new country, but I don't even wanna deal with the hassle of that. I think those two are, like, pretty fitting categories. We currently service companies in the production market space, skill based games, marketplaces, digital goods, subscriptions. Travel, don't know yet. I just had Yeah. I was just one conversation today. It's interesting, but it's

yeah. It's complicated. Yeah. Yeah. So when there's a lot of people dealing with a lot of manual expensive problems that could be automated Yeah. How does this practically take shape? So if I have a marketplace business or I'm selling a subscription, I want to ideally sell this thing to as many people as possible all around the world.

So I'll have a handful of these different partners that are really good at each of these specific regions, and Breeze is kind of, it sounds like, fighting for as much of that pie as possible globally. So you're sounds like maybe often playing nicer around others who are in certain markets, or you'll come in for, like, we're gonna do these four markets really well as you scale. Is that typically how it works? When companies go through hyper growth phase, they've, like, probably onboard one payments partner.

And they're just using them for everything. Yeah. So sometimes it could be like, they're selling internationally and they're like an EU business, let's say. Mhmm. And they're like, I want to move US transactions domestically.

And there's a lot of reasons why you want to do that. You have way better approval rates. Approval rates is a massive one which directly translated revenue. And then we look at companies like that and we're like, we can beat just your US rail. And they're really happy with that because then they don't have to open US bank accounts, set up US teams or whatever. Or it could be like vice versa, a EU launch, or it could be like Indonesia,

Australia, whatever it is. Are there typically downsides as well to not being located in that region? Like, you know, if I'm based in some country, I'm doing a lot of business in another country. I'm assuming, even for us, like, I'm invoicing a client or customer and they're, like, in France or something or Germany, it's like the idea of being an expert everywhere is really hard, especially when there are these, yeah, emerging use cases or new markets or things that are not your core.

Are you finding that Breeze is maybe coming in in that very helpful for growth or emerging markets where, yeah, there's like specialty that's needed and then maybe that's like a wedge to grow the business, like land and expand kind of? Yeah. So we currently primarily focus on digital goods. Digital goods themselves have lots of different sales taxes implications around the world. How do you define the in the payment code of digital goods? You have It could be like a subscription.

It could be like an in game digital good. NFTs are technically NFTs are technically

that kind of where that wedge came from? Yeah. That started there. But digital goods themselves have sales taxes. Like, even in The US Yeah. Different states have different sales taxes and you're supposed to remit that. Okay. And then many companies don't wanna do that. Actually, funny enough, Apple's, like, in app purchase product does that globally. Apple themselves is a giant merchant of record but they charge you 30% for it.

So a lot of our pitch is actually like, hey, now that Apple has opened up the floodgates, now you can just use Breeze and you don't have to use Apple and we're way cheaper than that. Mhmm. But it's like local tax remittances, those operations are also not easy. But even like opening bank accounts, maintaining treasury, those are all operational toil things. Yeah. Yeah. Is a markdown file a digital good? If we were to sell access

to let's just say, for example, we wanted to package this episode that we're having right here as insight packet for sale, like a micro payment. Yeah. An agent can instead of a human listening to this, an agent can come in, and they can get all of the insight they need in a beautiful package with a bow on it. And let's just say that we wanna charge

50¢ for that. Yeah. Would that be technically that sounds like a digital sale, like a digital good sale. Right? We generally think of it as a digital good. Within digital good, it could be like an online course. Yeah. So that's like a very specific definition of it. Online courses are actually just booming in general. Like, or like fitness classes. We have all sorts of versions like that.

I don't understand why there's a million people paying for Pilates classes online, but I they're paying for Pilates classes online. So, like, ICPs here would be, like, Hormozi's

school platform, WAP. These big, like that's where a ton of opportunity is. Yeah. There is a ton of opportunity there. There are a lot of new creators selling all kinds of things online. Yeah. Yeah. Okay. As we close here, when you think about opportunity, we're now at year four of the business. Is that right? Yes. Almost five? No. Almost four. Almost four. We're not even four yet. When you think about four years ahead, doubling the length of the company, what does success

look like? When you think about the future and you're honest with yourself and your team of like, this is what we're trying to achieve here, like, does that look like for you guys? So I do think four years is too long of a time frame Yeah. Given how much AI is moving. Yeah. I think we used to plan everything on a three month basis, and then now it's, like, every one month. Actually, every week, pretty much.

I was joking with my wife. I was like I was like, yesterday, I was like, you know, two or three of these spots open. And she's like, you don't you have a whole studio booked and you have only a couple founders booked for this thing. Like, you don't get it. Like, founders move, like, every twenty four hours. As long as you ask them, like, a day or two before, they'll say yes or no. It's just the way, like, people, founders and teams, they just they're so focused on, like, the next couple days or a week. Yeah.

But how do you yeah. You gotta balance that too.

Yeah. Like, the most exciting thing that is happening to this world right now is definitely AI. And the way we're thinking about AI is, like, the reason why enterprises love Breeze is because we're very reactive, we're very responsive, we go build a feature, we go launch a new market, we go lift the approval rate, whatever it is. They tell us a pain point and we're like, we can do it. Mhmm. Right? I wanna scale that with AI.

Imagine that you can go from ingesting like a sales conversation all the way down to an agent actually building a feature, deploying that into production. Give it a compliance thing or legal thing, go escalate to the right teams or partnerships, whatever, and then get that unblocked as quickly as possible. That's our current goal right now and we call that generally Breezes Intelligence. New hey. I wrote this doc internally. Say it again? Breeze as Intelligence.

I created a new doc of the company, and I was like, guys, this is the future of AI. This is the future where we're going. We need to build our whole stack to make it ready for this. Yeah. I know. I think I missed this in our some of the things we wanted to cover. So the adoption of some of these things internally.

Yeah. Not only is this, like, topic of discussion macro external, like agents are going around. They're gonna wanna transact, but these are also the primitives here are very powerful internally. Yes. Let's expose that a little bit more. When you think about building the business itself or the team itself and how you think about

growth, optimization, scale, how do you even make sense of that right now? Like, do you define Breeze's intelligence? Yeah. Like, January of this year. It's crazy to think that that was just January. In January of this year, I think Opus four point six just launched. Everyone came back from their holidays. I was living in San Francisco. I swear if you had a graph of how many hours people slept in San Francisco, it would have dropped off a cliff. Because of AI. Yeah. Like, because of Open's 4.6.

And then I think OpenClaw basically launched close to, like, the end of that month in January, I got my hands on that. And the second I did that, I slept for three hours every day of that week. And it was because I was like, there's so many applications I can build. That's beneficial to Breeze. There are all these things I wanna do, and we're, like, officially entering a new era. What was so magical about that? Because I I think a lot of people felt that, but what was it in your experience that was so magical there?

It was the first time I really felt like I had a team of people working for me, a team of agents. Basically, I provisioned all sorts of API keys. These are all personal things, not corporate things. Like, my security person blocked me from that. But, like, personal hobby projects, beneficial to the company, externals to the company, but if I were an external vendor, could sell the Super Ease, basically. But, like, I would maybe you could. Yeah. Yeah. Yeah.

I had a funny story here is some vendor, which I cannot name, was trying to pitch me a product and I was like, okay, I'll go Vibe I Code was gonna build that. Yeah. No. No. Because they they gave me a very offensive pricing arrangement and I was like, come on. Like, I can't pay that. I was like, get Vibe Code to your application. I think we literally went back to them and went like, our CEO Vibe Coded your platform. What was without naming who it was, what was the actual function they were selling?

What did you build? I if I if I talk about that, it be Too specific. It's it's a piece of software where, like, you can ingest some sort of information and then be able to build it anyways. Yeah. But I think, essentially, OpenClaw had this capability of deploying many different agents and had memory. Like, agents weren't, like, DOM anymore. Like, agents remembered what was happening, the context that they were given.

And so you could have one agent do something very specific, another agent do something very specific. And then I had a whole Slack channel to monitor what every agent was doing. Mhmm. It was the first time it really felt like a team. And I was like, oh, there are all these applications I wanna build. Previously, I had to go human configure certain things. Even on Vercel, like, human configure certain, like, deployments or go on some API platform, configure those things, and it would take me hours. Yeah.

And now it's done in, like, minutes. I just had to do the setup once, and then it was all done. Why is this so powerful for the conversation that we're having around the group of people that you're kind of selling into, the hundred, hundred and fifty person fraud, chargeback, compliance. Like, bring us a little bit into why having these agents work for you in this way is so powerful in that context. How far can you get? That's related to like Breezell's Intelligence.

The thesis of Breezell's Intelligence is a customer, whether it's an end user or a business, can come in and complain about something. Let's say a business is like, I wanna launch in x country, or a customer is like, why is my card getting declined and whatever? We would be able to go and figure out that information, that data, or sometimes the code change or whatever it is, some middle process that's all automatically done internally, and then we'd be able to give them an output quickly. Right?

So to an end user, it could be, oh, your card is being declined for insufficient funds. Why don't you try with a different card? Which actually is, like, really massive conversion rate uplift because some people just don't realize their debit card doesn't have money. Really dumb example. It's kinda like I am not a guy who's into cars or knows how they work or anything. When I get a check engine light, I'm like,

we gotta take this thing in. Like, I'm not gonna go try to fix this. My brother has one of those very simple devices, which I thought was magical, but apparently, they're very common where you just stick it in the car, and then it'll give you a code. And then that code will tell you exactly what's going on that, essentially, the guys at the auto shop do anyway. It's kinda like that. Typical payment denied or whatever. It's like, no. We can actually go in. We can see exactly what's happening.

We can identify specifically what's going on and then address it, deal with that in a very programmatic fast way. Yep. Instead of taking my car to the shop, waiting, and, like, being pissed. It's Yeah. Such a more delightful experience. Yeah. And the the same is true for business. When businesses tell you, hey, can you fix this thing? It could be something very minute. It's like somehow we just don't support in our API docs. And they're like, can you support this? And then we're like, okay, we're doing it.

And then currently, a human goes and prompts an agent to do it. I want that entire process to be frictionless. Yeah. Where is that exchange happening? Is that, like, on the API docs in some sort of, like, chat? No. It's Slack. So usually on Slack, they're like some engineer client. Yeah. Yeah. They could be like, oh, I want to pass the payment fee to the end customer.

And we're like, oh, for this specific rail, we don't support that at the moment. But an agent could code that very easily, and we can have a human to review that. But currently, it's a human takes that and then goes and prompts and then create a PR, and then a human reviews that and they get back. And they already think we're like really fast. Over like, we can be get this done by end of week. And they were like, wow, you guys are so responsive. I want that to be way faster. Yeah.

Because I generally think that the biggest bottleneck to like growth is your roadmap itself. And you can keep disrupting that if you can implement as much AI as possible. Wow. Yeah. AI is

modernizing, changing optimistically. Cynically, it's displacing. It's hurting. Like, there's a like, all of these haters as well. I feel like there are lot of people too who are probably in roles that are like, you know, I'm thinking about the hundred, hundred and fifty person payments operation. And, you know, I'll speak for my own perspective. Like, with some of these tools, you're not gonna need all of these people anymore, and we're starting to see that.

If you were to give advice to that organization or someone in that organization on how to stay alive and stay relevant and stay employable and valuable, specifically in the bucket of this payments operations fraud compliance. What would be your advice to them in order to really, like, continue to bring value to the organization? How do you see that? Two things. So internally,

we have ops people when, like, Millie is on a Slack channel deploying all these applications. The ops person was apparently freaking out going, like, my job is gonna be replaced by AI. Millie's doing all these things. My ops were like, internal ops people. Like, it's like, my job is gonna be replaced by AI and I heard from our our I'm afraid of that. Yeah. It's a very big fear, I think. Oh.

And then I think what ended up happening was I was like, look, we had a month long AI hackathon where it was like, there was a non injuring track and an injuring track. And then the non injuring track was think of things you're doing, like, manually and just automate that. And then we just kept pushing people to use AI and then they realized that they freed themselves up to do a lot more things. So it's like, oh, previously I had to go manually click this and that and that and that.

And then they went and built an agent to automate that process. And then that meant that they had more free time and they can do more things and they're now way more efficient. Right? So that's how we've been pushing that internally and that's how we approach some of these conversations which is, hey, you don't want payments to be your bottleneck. We can make that easier and you can free yourself up to do many other things that you can optimize on. Yeah. And that's how we generally think about work and AI too.

Like, I think the shape of work is changing. Engineers these days are sitting there prompting. Right? Engineers are no longer writing code. Jobs are just, like, changing shape. And it sounds like being

open, adaptive, using these tools, being aware of where pain is or where optimization could happen, and then building that muscle up as quickly as possible so that you're firing yourself first in some ways. What did we miss around your story, Breeze, what you're building, where you're going? I know we covered so many different pieces of this puzzle. Is are there any pieces that you felt like we haven't covered that deserve some spotlight or flowers that we've missed? I think we did a very thorough coverage. Good.

I think you probably are. This is your first, like, long form interview. How'd we do? Was it okay? Yeah. It was okay. It feels like it's way longer than I expected. You didn't spill too many beans. Well, cool, Millie. Thank you for this. Excited to continue to be in relationship with you, see this awesome journey unfold. And where can people find you? Yeah. What's the best place to learn more about Breeze? I know that you have a new website coming soon. Hopefully, by the time this episode comes out, we'll see it.

New website coming soon. I am posting a lot on LinkedIn. Would welcome

anyone to engage. I'm trying to scale up x. It's like a different challenge. It is a different beast. Yeah.

I see a lot of AI slop content on x lately. It's getting better. It's getting better? It's getting better. The one, x has gotten so much better. We originally started the business on LinkedIn, and that was, like, really where I felt like our community and following was with Zach and myself. And then when we launched the live show, like, x is so much better for video. Mhmm. And so most of the visibility in, like, this conversation is gonna do so much better on x because that's where people are mostly consuming video.

Same with YouTube. But LinkedIn is professional, and it's kinda like where you build professional relationships and network, which is everything we've talked about. The people who are actually gonna care are professionals probably listening for their jobs. It is a weird tension because LinkedIn is really bad for certain things, but it's also really good for others. X is what they have introduced, though, is the dislike button. Oh, really? I didn't see that. If you see a bot or comment or something Oh.

You can dislike something and flag it as like, this is an AI thing or whatever. And so I have just been going on rampages with people because you see these things all the time. Right? We'll we'll pause here. We don't need to waste people's stuff. Millie, thank you. This was fun. Yeah. Thank you so much for having me. You got it.