Inside Reap's Stablecoin-Powered Card Infrastructure

WEBVTT
00:00:03.040 --> 00:00:13.919
Welcome to Mint Condition, where we go beyond the headlines to truly understand how payments companies and financial institutions are actually leveraging crypto.
00:00:13.919 --> 00:00:20.559
Today's episode is about Reap, a stablecoin native card issuer built in Hong Kong in 2018.
00:00:20.559 --> 00:00:24.160
If that sounds like a modest start, the figures definitely aren't.
00:00:24.160 --> 00:00:30.320
Firstly, they have direct visa principal membership in three separate jurisdictions.
00:00:30.320 --> 00:00:39.520
Secondly, they've built a stablecoin collateralized credit facility running underneath a Visa bin instead of a debit rail.
00:00:39.520 --> 00:00:48.240
And most recently, they received a definitive agreement to be acquired by Kraken's parent company, Payward.
00:00:48.240 --> 00:00:52.240
My guest today is Harris, head of product at Reap.
00:00:52.240 --> 00:01:02.479
Harris started his career at MasterCard before moving on to Visa and had significant experience building crypto-native solutions globally.
00:01:02.479 --> 00:01:08.560
Just before he joined Reap, most notably, he was building international products at Coinbase.
00:01:08.560 --> 00:01:13.519
This conversation with Harris gets into the actual mechanics behind the headline.
00:01:13.519 --> 00:01:20.400
How a collateral engine turns stable coins into spending power without ever touching the payment rails itself?
00:01:20.400 --> 00:01:28.480
Why interchange economics change when a card is postpaid and collateralized rather than prepaid?
00:01:28.480 --> 00:01:36.719
And why Reap built for Asia's regulatory fragmentation before most Western issuers had a stablecoin strategy at all?
00:01:36.719 --> 00:01:41.840
We also spend real time on something that doesn't usually get discussed enough.
00:01:41.840 --> 00:01:51.840
What it actually takes, technically, to let an AI agent hold a corporate card spending power without ever letting it hold the card number.
00:01:51.840 --> 00:01:58.799
And where the infrastructure genuinely stands today versus where the announcements make it sounds like it stands.
00:01:58.799 --> 00:02:02.480
Now, on to the interview with Harris.
00:02:03.519 --> 00:02:07.040
That's uh that's definitely the hot topic of the day, Agentic.
00:02:07.040 --> 00:02:08.879
Um thought you never laughed.
00:02:08.879 --> 00:02:14.319
Um so so that this is yeah, this is a great point, Ryan.
00:02:14.319 --> 00:02:24.639
And with agentic payments, the core problem we're solving is that AI agents they can't hold a card number.
00:02:24.639 --> 00:02:34.240
The moment a PAN lives inside an LLM's uh context window, you you you've got a PCI ESS uh nightmare.
00:02:34.240 --> 00:02:39.919
Um and agents can't complete 3DS challenges, right?
00:02:39.919 --> 00:02:45.039
Because if you're doing it via an app or an SMS, how can the agent see it?
00:02:45.039 --> 00:02:48.240
So transactions just get declined at checkout.
00:02:48.240 --> 00:03:12.960
So what we have built or what we are building towards at Reap is a flow where human does the one setup step where they sign a spending policy with a biometric pass key, something like to simplify it, this rule or policy would be this agent can spend up to $500 a month at software mergers.
00:03:12.960 --> 00:03:13.520
Right?
00:03:13.520 --> 00:03:21.280
That creates what we call a scope token where the agent then uses instead of a real card number.
00:03:21.280 --> 00:03:26.800
Now, that agent in this scenario never gets to see the card details.
00:03:26.800 --> 00:03:37.120
And because the authentication already happened at the setup point when the human setup the policy, then the checkout challenge gets suppressed, right?
00:03:37.120 --> 00:03:39.520
And the agent just transacts.
00:03:39.520 --> 00:03:48.400
And what we have done here is essentially pull the level or rather the trust ceiling much closer to the card holder.
00:03:48.400 --> 00:03:59.520
Um reap, to your point, is one of the very, very small number of issuers globally that has built the infrastructure to support this model.
00:03:59.520 --> 00:04:05.360
Um we're in the final stages of going um live and we're building towards that direction.
00:04:05.360 --> 00:04:12.479
The pipeline is is real, the demand has come to us faster, much faster than we have expected.
00:04:12.479 --> 00:04:14.479
This is not a fed.
00:04:14.479 --> 00:04:17.279
We are seeing it in the market.
00:04:17.680 --> 00:04:25.920
And one of the interesting things about agentic payments, right, is now there's a slightly different way to think also about compliance and risk modeling.
00:04:25.920 --> 00:04:35.519
Um part of that is let's say you may be giving your and your agent or an agent the ability to spend using a Reap card.
00:04:35.519 --> 00:04:47.040
But one of the interesting things I'm or one of the things I'm curious to understand how you manage is how do you manage rescinding that access right?
00:04:47.040 --> 00:04:56.720
Um, so for example, maybe an agent could be compromised or could hallucinate on spending decisions or act outside of intent.
00:04:56.720 --> 00:05:05.199
So what is the enforcement mechanism if an agent presents a valid credential for a transaction that violates the actual business intent?
00:05:05.199 --> 00:05:08.319
Um, but it's technically within the scope parameters.
00:05:08.720 --> 00:05:08.959
100%.
00:05:08.959 --> 00:05:11.360
I think those are very important questions.
00:05:11.360 --> 00:05:15.920
They're questions that we ask ourselves as we build towards uh this product.
00:05:15.920 --> 00:05:22.000
Um and as we, I guess at Reap, we're building, we're in the midst of building this product.
00:05:22.000 --> 00:05:26.000
And so those questions you ask are essentially the same questions that we have.
00:05:26.000 --> 00:05:28.079
Um we're gonna solve them.
00:05:28.079 --> 00:05:37.839
And once we launch, you'll be sure, I'll be sure to give you the alpha beta access to it so that you can be our early launch user.
00:05:38.160 --> 00:05:38.800
Let's go.
00:05:38.800 --> 00:05:40.079
I I want to do that.
00:05:40.079 --> 00:05:40.720
I'm ready.
00:05:40.720 --> 00:05:44.800
Um, I will be messaging you, Harris, if I if I'm not.
00:05:44.800 --> 00:05:50.160
Um, and everyone else on listening to this podcast heard it, so you definitely have to follow.
00:05:50.160 --> 00:05:55.120
Two problems are getting solved here, and it helps to hear them separately.
00:05:55.120 --> 00:05:58.319
The first is about where car data actually lives.
00:05:58.319 --> 00:06:05.120
Once a car number sits inside an AI model's memory, that's a genuine problem for PCI compliance.
00:06:05.120 --> 00:06:10.399
You can't control what a model retains, and you can't guarantee it gets deleted.
00:06:10.399 --> 00:06:13.680
So the safest fix is the simplest one.
00:06:13.680 --> 00:06:16.879
Never give the agent the real car number at all.
00:06:16.879 --> 00:06:21.920
What Harris calls a scope token is a stand-in credential.
00:06:21.920 --> 00:06:34.079
My own personal read on this, and this isn't what Harris said or what Reap had said outright either, is that it works a bit like the network tokens on Apple Pay and Google Pay, which are already in use.
00:06:34.079 --> 00:06:39.040
The agent gets a token, the real number stays locked away with Reap.
00:06:39.040 --> 00:06:45.439
The second problem is about authentication, and it's really a channel problem.
00:06:45.439 --> 00:06:51.120
A normal 3D secure check pushes a prompt to your phone and waits.
00:06:51.120 --> 00:06:54.720
An agent doesn't have a phone to receive that on.
00:06:54.720 --> 00:07:02.240
So REIT moves the authentication earlier to the moment the human sets up the spending policy using a passport.
00:07:02.240 --> 00:07:07.120
That one moment is what every later transaction gets checked against.
00:07:07.120 --> 00:07:10.480
Instead of triggering a fresh challenge each time.
00:07:10.480 --> 00:07:16.000
It's worth knowing this kind of infrastructure is already live in the market.
00:07:16.000 --> 00:07:23.199
Reap sits on Visa's own list of approved partners for agent-based payments across Asia Pacific.
00:07:23.199 --> 00:07:28.959
I wanted to actually go uh back one step, right?
00:07:28.959 --> 00:07:33.680
Um, to it's talking a bit about agentiate payments and card-based payments.
00:07:33.680 --> 00:07:40.639
Um, I'm curious, when you're building a product for agenti payments, what are some of the obviously there's a way in which you're building, right?
00:07:40.639 --> 00:07:46.319
Or you built your card networking system for stablecoin native or crypto native payments.
00:07:46.319 --> 00:07:52.560
What are some of the different assumptions or models that you have to work with when it comes to agenti-based payments?
00:07:52.560 --> 00:07:59.680
I'm curious as to how even thinking about a card-based product looks different in that world.
00:08:00.000 --> 00:08:01.199
That's a great question.
00:08:01.199 --> 00:08:27.839
And I and I think one of the I wouldn't say challenge, but unique aspects of building agentic payments is that you essentially have to work with multiple partners and stakeholders within the ecosystem who have a different interpretation of payment protocols or rather agentic payment protocols.
00:08:27.839 --> 00:08:57.279
And so I think the challenge or the opportunity really for the next innovation within agentic payments is how we can find a way to build a unified layer that makes agentic payments a single interface, or rather, provide a single interface for clients and businesses to essentially integrate agentic payment capabilities into their products.
00:08:57.279 --> 00:09:04.960
I think that's going to be key in order to ensure that agentic payments um goes mainstream.
00:09:05.360 --> 00:09:17.919
As part of that, I imagine, right, for example, when you're when a platform normally launches a car program, so outside of the agency payment space, right, um there'll be certain controls that will sit with them, certain controls that will sit with you.
00:09:17.919 --> 00:09:25.919
Um does that vary or radically change when you're doing it for agentic payments, or is are some of those controls really similar and aligned?
00:09:25.919 --> 00:09:36.080
So for example, I don't know if in a traditional um payment, in your traditional payment scheme, traditional crypto is not that traditional, but traditional relative to agentic payments, right?
00:09:36.080 --> 00:09:42.159
Um you I imagine maybe the business has control over like KYB, right?
00:09:42.159 --> 00:09:46.960
And how they're thinking about KYB and their users to some extent, right?
00:09:46.960 --> 00:09:50.480
Um you all have certain parameters that they have to hit.
00:09:50.480 --> 00:10:00.720
Um but does that differ heavily when you're thinking about the parameters and who has control of those parameters for agentic payments?
00:10:00.720 --> 00:10:02.799
I'm curious as to what that actually looks like.
00:10:03.200 --> 00:10:04.080
Great question.
00:10:04.080 --> 00:10:39.120
So the way I see it, or the way we have been approaching it at Reap, is actually bringing ourselves closer to the networks that are building the core infrastructure of agentic payments so that we can one ride on the existing controls that are already in place and we don't have to essentially rebuild them, but two, establish the best practices that will be needed in order for agentic payments to scale to the scale that non-agentic payments have.
00:10:39.120 --> 00:10:40.879
And I think that will be key.
00:10:40.879 --> 00:10:54.480
It's more of a collaboration, it's more of a partnership, it's more of being a design partner to all the stakeholders within the ecosystem so that REIT can essentially be the one-stop shop to enable and unlock agentic payments.
00:10:54.799 --> 00:10:55.279
Got it.
00:10:55.279 --> 00:11:05.360
And I guess how I guess as part of that, we'll be thinking about how you manage, let's say, multi-agent systems where maybe there's a sub-agent or contractor agent or whatever, right?
00:11:05.360 --> 00:11:09.600
Um, and that's one of the interesting nuances about agentic payments.
00:11:09.600 --> 00:11:17.039
Um, how do you manage all these different systems that maybe it's one agent with one platform, but another agent that's been managed by another platform as well.
00:11:17.039 --> 00:11:21.600
Um, and yeah, we're still early on that front.
00:11:21.600 --> 00:11:26.879
Harris causes different interpretations of agentich payment protocols.
00:11:26.879 --> 00:11:34.240
But what's actually happening is two separate systems built by almost two separate sets of companies.
00:11:34.240 --> 00:11:38.720
And a player like Reap has to sit underneath both at once.
00:11:38.720 --> 00:11:40.799
Let's start with identity.
00:11:40.799 --> 00:11:47.120
This layer answers one question: Is it a legitimate agent or is it a bot?
00:11:47.120 --> 00:11:52.159
Visa's answer to this is called trusted agent protocol.
00:11:52.159 --> 00:11:56.159
It launched last October, built with Cloudflare.
00:11:56.159 --> 00:12:03.840
It signs an agent's identity into every transaction, checked in real time and gates a directory that Visa maintains.
00:12:03.840 --> 00:12:09.360
This sits on the merchant side, checking legitimacy before checkout even starts.
00:12:09.360 --> 00:12:16.799
A different job from what Harris described a moment ago, which happens on the issuer's side, inside the payments itself.
00:12:16.799 --> 00:12:24.720
And trusted payment protocol is really just one piece of something much bigger, that Visa is building as well.
00:12:24.720 --> 00:12:28.639
They call the whole thing Visa Intelligent Commons.
00:12:28.639 --> 00:12:37.600
It started rolling out to banks and issuers in Europe back in March, then expanded into APAC and Latam at the end of April.
00:12:37.600 --> 00:12:51.600
The clear signal is that Visa wants to be the trust layer everyone roots through, no matter which agent or which network is actually moving the money.
00:12:51.600 --> 00:12:55.759
Now, the second system to really focus on is to do with settlement.
00:12:55.759 --> 00:12:58.559
So, i.e., actually moving the money.
00:12:58.559 --> 00:13:05.120
And that's a different race entirely with protocols such as X402 gaining significant traction.
00:13:05.120 --> 00:13:07.679
Now back to the interview.
00:13:07.679 --> 00:13:13.200
I'd love to understand more about how the collateral engine actually works mechanically.
00:13:13.200 --> 00:13:19.840
So who holds it, um, how limits are set, what happens at settlement, what does this actually mean?
00:13:19.840 --> 00:13:29.440
And, you know, of course, for many payment execs and big institutions, they're very familiar with how cards work in a traditional sense, just with feat pure fiat rails.
00:13:29.440 --> 00:13:37.200
I'm intrigued for you to unpack what that looks like on-chain and how um Reap approaches that as well, specifically.
00:13:38.080 --> 00:13:39.039
Great question.
00:13:39.039 --> 00:13:42.559
Um essentially, what's our secret source?
00:13:42.559 --> 00:13:53.519
Um, the easiest way to think about it is that the collateral engine is what transforms digital assets into usable spending power.
00:13:53.519 --> 00:14:04.240
Now, a customer begins by depositing stable coins such as USDC and USDT, which act as a collateral for the program, right?
00:14:04.240 --> 00:14:18.559
And once those funds are available, our platform can essentially allocate spending limits against that collateral, allowing cardholders to transact through the existing acceptance network just like any other commercial card.
00:14:18.559 --> 00:14:24.000
From the card holder's perspective, though, nothing feels different.
00:14:24.000 --> 00:14:33.200
They tap their card, the transaction is authorized, and the merchant receives payment through the existing cardrails.
00:14:33.200 --> 00:14:40.320
What is different though is that everything is happening behind the scenes.
00:14:40.320 --> 00:14:52.879
Every transaction is continuously reconciled against the collateral balance so that businesses have a real-time view of the available spending, right?
00:14:52.879 --> 00:14:56.000
The settlement obligations and treasury positions.
00:14:56.000 --> 00:15:09.440
So because the collateral is uh is already in place, businesses don't have to constantly move liquidity and working capital between banks, exchanges, and card providers to keep the programs funded.
00:15:09.440 --> 00:15:11.600
That's the real innovation.
00:15:11.600 --> 00:15:19.519
Um the collateral isn't just there for security, it becomes the funding engine for the entire card program, really.
00:15:19.519 --> 00:15:24.000
So, more broadly, we see this as part of a much bigger shift.
00:15:24.000 --> 00:15:30.960
Um, traditionally, businesses have had one system managing treasury and another managing payments.
00:15:30.960 --> 00:15:49.360
Our approach starts with bringing those two roles together, allowing digital assets to move beyond being something that businesses simply hold and instead become productive working capital that can directly fund day-to-day operations.
00:15:49.360 --> 00:15:54.559
Essentially, we're not trying to reinvent how people pay.
00:15:54.559 --> 00:15:58.080
We're reinventing how businesses fund payments.
00:15:58.080 --> 00:16:07.440
Um, and and the user experience needs to stay familiar, the infrastructure underneath becomes significantly uh more efficient.
00:16:10.320 --> 00:16:12.240
So I'm curious, right?
00:16:12.240 --> 00:16:26.000
Because one of the interesting things about Reap cards, right, is that um whereas most other cards are prepaid, right, from my understanding, Reap is more of a postpaid card.
00:16:26.000 --> 00:16:29.600
And so how does that work from a collateral and risk management perspective?
00:16:29.600 --> 00:16:37.440
Um, it's a different structure, and arguably it's um more akin to what you would see in a lot of traditional card rails, right?
00:16:37.440 --> 00:16:45.840
It's you know, we don't necessarily need to provide collateral on average to be able to leverage a card in most, at least developed economies.
00:16:45.840 --> 00:16:49.759
And so, how does the prepaid versus postpaid model work?
00:16:49.759 --> 00:16:55.360
And how are you actually able to do this leverage in crypto when many other players don't have this model?
00:16:55.840 --> 00:16:57.200
Absolutely spot on.
00:16:57.200 --> 00:17:01.600
Most crypto cards today are essentially prepaid cards, right?
00:17:01.600 --> 00:17:03.440
Prepaid debit cards.
00:17:03.440 --> 00:17:09.759
Every time you want to spend, you first have to sell or convert your crypto to fund the transaction.
00:17:09.759 --> 00:17:14.160
Your assets and your spending are very much tightly coupled.
00:17:14.160 --> 00:17:18.240
And this is where Reap differentiates ourselves.
00:17:18.240 --> 00:17:19.839
We took a different approach.
00:17:19.839 --> 00:17:24.960
With Reap, stable coins are used as collateral rather than as the payment itself.
00:17:24.960 --> 00:17:32.720
And that means businesses can lock value on the platform for spending against credit lines through the existing card networks.
00:17:32.720 --> 00:17:34.880
Now, why is that important?
00:17:34.880 --> 00:17:39.119
I think the first is that it creates a much better payment experience, right?
00:17:39.119 --> 00:17:46.720
Credit cards have a higher merchant acceptance and better interchange as compared to prepaid and debit cards.
00:17:46.720 --> 00:17:50.160
And in general, much stronger economics.
00:17:50.160 --> 00:17:57.119
Businesses don't have to change how they pay, only how the program is funded.
00:17:57.119 --> 00:18:01.440
Now, the second bit is that it separates treasury from spending.
00:18:01.440 --> 00:18:11.200
A business can essentially decide what assets it wants to hold as collateral while independently managing how and when it spends.
00:18:11.200 --> 00:18:15.359
Um, and that opens up an entirely new operating model.
00:18:15.359 --> 00:18:25.519
Today that might be stable coins, tomorrow it could be tokenized uh treasury, yield-bearing assets or other forms of uh digital collateral.
00:18:25.519 --> 00:18:36.240
Because the collateral and the spending engine are decoupled, the funding model becomes much more flexible over time and can scale with every card program.
00:18:36.240 --> 00:18:42.319
Now, ultimately, we didn't set up to build another crypto card.
00:18:42.319 --> 00:18:46.799
We built a programmable funding layer for commercial card programs.
00:18:46.799 --> 00:18:51.599
And that's why many of our customers aren't actually looking for a crypto product.
00:18:51.599 --> 00:19:01.200
They're looking for a faster way to launch a global card program without tying up working capital or rebuilding any financial infrastructure.
00:19:01.200 --> 00:19:21.200
I think a great way to exemplify this is one of the examples is a neobank that we work with that wanted to launch a stable coin back card program rather than spending, you know, over a year building issuing infrastructure, compliance, fraud systems, and integrating with payment networks.
00:19:21.200 --> 00:19:29.759
They essentially launched with Reap in under three months, just using Reap's white label cards as a service platform.
00:19:29.759 --> 00:19:35.200
From their customers' perspective, it behaves like a normal payment card.
00:19:35.200 --> 00:19:46.880
Behind the scenes, stablecoins provide the collateral funding while Reap handles the card issuance, compliance, and settlement, and essentially the global payments infrastructure.
00:19:46.880 --> 00:19:50.160
And I think that's really the philosophy behind the product.
00:19:50.160 --> 00:19:56.720
We aren't changing how people pay, we're changing how payment programs are funded.
00:19:58.079 --> 00:19:59.200
I'm curious, right?
00:19:59.200 --> 00:20:12.799
So with, for example, the fact that you know, Reap and this is, you know, you you're it's a common, I guess, thread within the crypto industry, whether it comes to cards, whether it comes to lending protocols, right?
00:20:12.799 --> 00:20:17.759
Or borrow lend protocols, um, is the idea of needing collateral.
00:20:17.759 --> 00:20:23.039
For many in the traditional payments industry, right, their question is, well, why do we need collateral?
00:20:23.039 --> 00:20:25.200
Um, what does that look like?
00:20:25.200 --> 00:20:28.880
Um, what does that mean in the context of crypto?
00:20:28.880 --> 00:20:39.599
So of course it can be stable coins, of course it can be tokenized RWAs, but why is that collateralized model so fundamental and when it comes to cards in particular, obviously, um crypto native cards?
00:20:39.599 --> 00:20:41.839
I'm curious to get your thoughts on that.
00:20:43.119 --> 00:21:04.640
I think the collateral aspect essentially gives us the ability to expand our operating models across different types of card programs and cater to different personas of client portfolios as well, whether it's retail cards, commercial cards, Web3, or even Web2 card programs.
00:21:04.640 --> 00:21:11.279
And I think that's where the power and value of having a robust collateral engine uh comes in.
00:21:12.000 --> 00:21:16.799
Put this next to how a traditional secured business card actually works.
00:21:16.799 --> 00:21:23.759
In a standard sponsored bank model, the structure that most banking as a service providers actually run on.
00:21:23.759 --> 00:21:27.279
A fintech partners with a chartered bank.
00:21:27.279 --> 00:21:31.359
So a chartered bank being the lights of a cross river or a statement bank.
00:21:31.359 --> 00:21:38.640
And the bank holds the bid, they're the issue of record, and they carry the regulatory relationship with the network.
00:21:38.640 --> 00:21:51.759
The fintech is a program manager sitting on top, and interchange usually gets splits between them at a negotiated figure, somewhere between the 60 to 80% range.
00:21:51.759 --> 00:21:59.519
For a secured version of this product, collateral usually takes the form of cash held in an FDI C insured.
00:21:59.519 --> 00:22:03.200
Deposit account at the same bank that's issuing the card.
00:22:03.200 --> 00:22:10.240
The bank takes a security interest and a right of set off against that specific account.
00:22:10.240 --> 00:22:20.400
Because the collateral and issuers sit inside one regulated balance sheet in one currency, the bank doesn't need to worry about cost asset basis risk.
00:22:20.400 --> 00:22:29.440
So LT fee is essentially 100% dollar for dollar and the risk model is closer to hold a CD, don't touch it, than active margining.
00:22:29.440 --> 00:22:32.720
But that structure has a hard dependency baked in.
00:22:32.720 --> 00:22:40.640
The business has to be bankable at the institution, in that currency, in that jurisdiction, before any of this works.
00:22:40.640 --> 00:22:44.240
It's exactly the population Reap is targeting.
00:22:44.240 --> 00:22:53.440
Those crypto native businesses, DeFi protocols, crypto exchanges, and separately, businesses in markets with underdeveloped banking.
00:22:53.440 --> 00:23:05.519
Because a chartered deposit-taking institution has real constraints around what counts as an acceptable deposit relationship.
00:23:05.519 --> 00:23:20.240
What makes Reap structure significant is that being a visa principal member directly rather than program manager under a sponsor bank means the collateral doesn't have to sit inside a chartered deposit account at all.
00:23:20.240 --> 00:23:33.119
It can sit with a third-party digital asset custodian denominated in USDC or USDT and Reap holds a security interest against that custodial position instead of a bank deposit.
00:23:33.119 --> 00:23:38.160
That decouples card issuance from local bank chartering entirely.
00:23:38.160 --> 00:23:50.400
A business anywhere can post stablecoin collateral once and get a card issued in USD or Hong Kong dollars, etc., without opening a local currency account in each market operates in.
00:23:50.400 --> 00:23:59.920
It also means Reap keeps a full interchange, or at least most of it, rather than splitting it with a sponsor because there is no sponsor.
00:23:59.920 --> 00:24:01.519
But there is a trade-off.
00:24:01.519 --> 00:24:07.680
And the trade-off is that Reap inherits a harder risk problem than the traditional model ever had to solve.
00:24:07.680 --> 00:24:12.720
A bank holding cash collateral in its own ledger has no repricing to do.
00:24:12.720 --> 00:24:19.359
The deposit and the credit line are the same currency, same institution, same instant.
00:24:19.359 --> 00:24:37.119
Reap's collateral is separate, custodied, technically distinct assets from the credit line it backs, which is why the engine has to continuously reconcile spend against collateral value in real time rather than relying on a static deposit relationship.
00:24:37.119 --> 00:24:39.519
Now, back to the interview.
00:24:39.519 --> 00:24:44.240
Going back slightly, right, there's something that you mentioned that actually triggered a thought, right?
00:24:44.240 --> 00:24:55.920
Um, I'm curious, right, going back to the point about collateral, and you know, your model is really interesting because of course it helps to manage time and floats, right?
00:24:55.920 --> 00:25:01.759
Which sometimes can be a challenge, a working capital challenge, right, for businesses that have a card scheme.
00:25:01.759 --> 00:25:13.359
But one of the equal challenges, and this is not just you, this is the crypto card industry more broadly, is that of course, when you have collateral, that's capital that is just locked up, right?
00:25:13.359 --> 00:25:16.480
That you have to have ready sitting idle.
00:25:16.480 --> 00:25:23.359
And especially if you're a business that your card program is continuing to scale, your collateral requirements will also scale.
00:25:23.359 --> 00:25:32.640
So, how does therefore card crypto native card programs really work at true large scale, given this collateral requirement?
00:25:32.640 --> 00:25:39.680
Is it something in the system or in the way in which these products can be built to help make that more manageable and more affordable for these businesses?
00:25:39.680 --> 00:25:46.720
Because not all businesses have that capital or at what level of working capital to just have sit parking idle.
00:25:46.720 --> 00:25:48.079
100%.
00:25:48.640 --> 00:25:49.119
100%.
00:25:49.119 --> 00:26:02.160
And so today, when we look at all these um working capital and car programs, there is this artificial uh separation between treasury management and payments, right?
00:26:02.160 --> 00:26:12.160
Finance teams, they they hold cash in one place to earn yield and then move it somewhere else when they need to pay suppliers, find employees, or make cross-border payments.
00:26:12.160 --> 00:26:17.359
All that that movement creates is idle capital, right?
00:26:17.359 --> 00:26:22.559
Operational friction and a lot of unnecessary settlement delays.
00:26:22.559 --> 00:26:37.839
I think what stablecoins and tokenized assets make possible is a different operating model where the same pool of working capital can simultaneously remain productive and be instantly deployable for payments.
00:26:37.839 --> 00:26:42.240
And that's a very powerful value proposition.
00:26:42.240 --> 00:26:56.000
Imagine a business treasury where idle balances are invested in tokenized treasury funds or other regulated deal-bearing assets, generating returns throughout the day.
00:26:56.000 --> 00:27:14.000
Then the moment an employee swipes a corporate card or a supplier payment is initiated, right, that that liquidity is made available instantly without finance teams manually moving money between multiple banks, wallets, and investment accounts.
00:27:14.000 --> 00:27:18.079
To me, that that is a real opportunity.
00:27:18.079 --> 00:27:32.240
The innovation isn't simply putting yield like on a card, it's it's turning treasury into programmable infrastructure where liquidity is always working, but it's always available.
00:27:32.240 --> 00:27:38.720
There's this that's very much aligned with the direction that we're building towards a Reap.
00:27:38.720 --> 00:27:45.519
We have recently, very recently, announced our integration with USYC, right?
00:27:45.519 --> 00:27:52.640
Which gives businesses access to tokenized uh yield-bearing treasury exposure directly with REIT direct.
00:27:52.640 --> 00:28:01.359
Uh, it's an important step because it extends our platform beyond payments and expense management into treasury optimization.
00:28:01.359 --> 00:28:22.160
Right, and our vision is to give finance teams a single operating platform where they can manage liquidity, earn yields on idle capital, issue corporate cards, make cross-border payments, and settle it all using either fiat or stablecoins, all without stitching together multiple providers.
00:28:22.160 --> 00:28:27.200
I don't think CFOs wake up in the morning wanting DeFi.
00:28:27.200 --> 00:28:28.480
You know, I want DeFi.
00:28:28.480 --> 00:28:31.680
They want better cash management.
00:28:31.680 --> 00:28:48.079
And and if if blockchain can help their working capital earn more, move faster, remain fully auditable and controllable, I think then the underlying technology becomes most um invisible, and that's when you know it's working.
00:28:49.039 --> 00:29:05.119
That's the infrastructure that Reap has actually built underneath a simple card product, and it's why extending the eligible schedule from stable coins to a tokenized money market fund is a meaningfully different technical rift than itself.
00:29:05.119 --> 00:29:19.279
USDC and USDT or any US dollar stable coin is fundamentally priced against a one-to-one redemption check, i.e., is the peg holding yes or no?
00:29:19.279 --> 00:29:29.599
In contrast, a money market fund that is tokenized, well, its value is enough that accrues and needs a genuine price of fee, not just a peg check.
00:29:29.599 --> 00:29:43.759
The haircut that Reap decides to any tokenized money market fund should, by the logic of every other margin lending system that prices T-fill-based collateral be smaller than the haircut on a more volatile asset.
00:29:43.759 --> 00:30:00.480
But that only holds if the Oracle feeding that nav into the collateral engine is reliable and frequent enough to catch a redemption gate or a pricing dislocation before it becomes Reap's problem instead of the fight.
00:30:00.480 --> 00:30:08.799
Now, Reap has thought about this at length and we'll be exploring this and more shortly.
00:30:08.799 --> 00:30:11.359
So let's go back to the interview.
00:30:11.359 --> 00:30:17.359
So most corporate card programs, right, they make money on interchange.
00:30:17.359 --> 00:30:23.200
Um, your model, of course, layers interchange with FX spread and subscription fees.
00:30:23.200 --> 00:30:30.720
Um, how do these three revenue lines interact and which one scales fastest as volumes grow?
00:30:31.440 --> 00:30:32.640
Great question.
00:30:32.640 --> 00:30:40.799
Um I I don't think I don't think stable coins disintermediate cart networks overnight.
00:30:40.799 --> 00:30:44.960
I think they they actually force the economics to to evolve.
00:30:44.960 --> 00:30:52.640
Um cart networks have, you know, when we look back in time, right, have adapted through every major real shift in payments.
00:30:52.640 --> 00:31:00.240
Online commerce, uh mobile wallets, real-time payments, open banking, and and now stable coins.
00:31:00.240 --> 00:31:02.000
The reason is simple.
00:31:02.000 --> 00:31:05.680
The acceptance network is extremely hard to replicate.
00:31:05.680 --> 00:31:10.480
Hundreds of millions of merchants already accept these networks, right?
00:31:10.480 --> 00:31:15.200
And that value is very hard, incredibly hard to replicate.
00:31:15.200 --> 00:31:30.000
So interchange isn't only a settlement fee, it actually also supports the broader card ecosystem with network access, authorizations, dispute handlings, all those good stuff.
00:31:30.000 --> 00:31:39.279
Um and all these things on-chain rails do not yet replicate at a global consumer scale.
00:31:39.279 --> 00:31:46.960
So my view, yeah, my view is that interchange survived, but it becomes more new nuanced.
00:31:46.960 --> 00:31:53.200
In some corridors and in some use cases, stablecoin settlement may compress parts of the economics.
00:31:53.200 --> 00:32:04.559
In others, the value of instant funding, global acceptance, risk uh management, and regulated infrastructure may justify a different fee model altogether.
00:32:04.559 --> 00:32:14.000
So for Reap, that is exactly why our strategy is not um dependent on one revenue line or one rail.
00:32:14.000 --> 00:32:20.559
Our role is to sit at the intersection of um stablecoin infra and the cart network infra.
00:32:20.559 --> 00:32:36.640
So our visa principal membership in Hong Kong and Mexico are important because they allow us to participate uh directly in how these new models develop rather than simply reacting from the outside.
00:32:36.640 --> 00:32:37.759
Right?
00:32:37.759 --> 00:32:44.000
So over time, I think winners will be companies that can essentially bridge both worlds.
00:32:44.000 --> 00:32:55.599
The real the reliability, the compliance, the acceptance of traditional card networks with the speed, programmability, and settlement efficiency of stablecoins.
00:32:55.599 --> 00:32:59.839
I think that is where Reap is very much uniquely positioned.
00:32:59.839 --> 00:33:02.960
We're not betting that one rail destroys the other.
00:33:02.960 --> 00:33:11.920
We're betting that the future of payments is going to be hybrid and that businesses will need infrastructure that can operate across both.
00:33:12.400 --> 00:33:22.640
Harris names three revenue lines, but what's really interesting is understanding what stablecoins enables or how stable coins transform these revenue lines.
00:33:22.640 --> 00:33:26.400
So let's double down on interchange in particular.
00:33:26.400 --> 00:33:32.559
In a normal card program, interchange revenue and settlement timing are locked together.
00:33:32.559 --> 00:33:36.640
The issuer collects interchange on authorization.
00:33:36.640 --> 00:33:40.799
Let's double down on interchange in particular.
00:33:40.799 --> 00:33:46.799
In a normal card program, interchange revenue and settlement timing are locked together.
00:33:46.799 --> 00:33:58.799
The issuer collects interchange on authorization, but funding that authorization actually settling with the acquirer and getting reimbursed runs on the card network's clearance cycle.
00:33:58.799 --> 00:34:05.119
So typically a T plus 1 or T plus 2 wire or ATH movement through correspondent banking.
00:34:05.119 --> 00:34:18.800
Reap can collect the same interchange on the authorization side, fully on Visa's Rails, while funding its own settlement obligations using stablecoin liquidity that moves in minutes rather than days.
00:34:18.800 --> 00:34:20.800
That's not a new fee.
00:34:20.800 --> 00:34:27.039
It's the same interchange revenue with a working capital drag underneath it, structurally shortened.
00:34:27.039 --> 00:34:38.400
Which matters because the funding gap is exactly the kind of locked up capital Paris was describing earlier as the industry's core scaling constraint.
00:34:38.400 --> 00:34:45.599
Reap is amongst the top tier when it comes to like cards within crypto, the crypto industry.
00:34:45.599 --> 00:34:50.239
Um, and I actually want obviously you are an Asian-based company.
00:34:50.239 --> 00:34:54.719
Um, and I wanted to touch start by touching upon that.
00:34:54.719 --> 00:35:04.639
So Asia now accounts for roughly 60% of global stablecoin payment volume, um, according to a recent study by McKinsey and Artemis.
00:35:04.639 --> 00:35:06.800
You were building in Hong Kong, right?
00:35:06.800 --> 00:35:12.559
Before most of the West had any clear regulatory framework, before the Genius Act, etc.
00:35:12.559 --> 00:35:20.719
Um, what did you see in Asia that made it structurally more ready for stablecoin payments than in the US or in Europe?
00:35:21.440 --> 00:35:23.440
That's uh that's a great question, Maya.
00:35:23.440 --> 00:35:29.920
I I actually think we had the opposite mindset to a lot of other companies.
00:35:29.920 --> 00:35:33.519
We didn't choose to build in Asia because it was easier.
00:35:33.519 --> 00:35:39.119
We actually chose to build here because it was actually harder and more complex.
00:35:39.119 --> 00:35:45.199
Because if you can solve payments in Asia, you practically can solve payments almost anywhere.
00:35:45.199 --> 00:35:54.719
So maybe what would be good is a quick introduction about Reap and what we do, and then the story will tell itself.
00:35:54.719 --> 00:36:02.239
So Reap is essentially a stablecoin native infrastructure that powers card programs and payments for businesses uh globally.
00:36:02.239 --> 00:36:21.199
What Reap is in a nutshell is that we connect stablecoin settlements with global banking rails and cart networks into a single platform, allowing fintechs, enterprises, neobanks, and financial institutions to launch payment card products without having to build the underlying infrastructure themselves.
00:36:21.199 --> 00:36:28.960
Um, and on the head of Product for Reap, so I take care of all the products that underlie this infrastructure that we have uh here at Reap.
00:36:28.960 --> 00:36:31.519
So back to why Asia.
00:36:31.519 --> 00:36:46.159
If we think about the reality of the region, you're dealing with dozens of currencies, fragmented banking systems, different regulatory regimes, varying payment rails, and cross-border settlements that you know can take days, right?
00:36:46.159 --> 00:36:50.639
It's one of the most operationally complex payment environments in the world.
00:36:50.639 --> 00:36:57.199
Now, that complexity has actually forced us to build differently from day one.
00:36:57.199 --> 00:37:11.280
So instead of designing, you know, for a single market or a single payment rail, we essentially built a platform that could operate across multiple currencies, jurisdictions, banking partners, and networks, right?
00:37:11.280 --> 00:37:16.960
And now even stablecoin settlements, all through that single infrastructure API layer.
00:37:16.960 --> 00:37:25.519
Um, Hong Kong was particularly strategic to do that because it's it's the world's largest FX hub, right?
00:37:25.519 --> 00:37:33.599
It's one of the few places where multiple major currencies settle efficiently and increasingly a global center for regulated digital assets.
00:37:33.599 --> 00:37:39.599
It sits at the intersection of traditional finance and kind of the next generation of financial infrastructure.
00:37:39.599 --> 00:37:55.440
What we saw early was that stable coins weren't simply a crypto innovation, they were a better settlement rail for one of Asia's biggest problems, which is moving money across borders quickly and efficiently.
00:37:55.440 --> 00:38:01.760
So, because of that insight, it has become increasingly relevant everywhere else.
00:38:01.760 --> 00:38:10.960
And so today we see that you know in the US, Europe, they are arriving at many of the same conclusions around stablecoin regulation and adoption.
00:38:10.960 --> 00:38:20.960
But because we started in Asia, and we started with the hard stuff, we have already spent years building the operational infrastructure that businesses actually need, right?
00:38:20.960 --> 00:38:25.360
The compliance, the treasury, the issuance, the FX, the cross-border settlement.
00:38:25.360 --> 00:38:31.039
So in hindsight, building in Asia wasn't just an early bet on stable coins.
00:38:31.039 --> 00:38:39.199
It was actually an early bet that the future payments would be multi-currency, multi-real, and global just by default.
00:38:39.199 --> 00:38:43.119
And that's exactly the world that we're operating in today.
00:38:43.760 --> 00:38:48.000
Here's a concrete version of the problem Harris is describing.
00:38:48.000 --> 00:38:56.960
A single currency ledger, the kind that most card programs start with, stores balances as one number in one denomination.
00:38:56.960 --> 00:39:04.960
And every downstream system, so authorization, special mode, reconciliation, etc., inherits that assumption.
00:39:04.960 --> 00:39:09.199
Adding a second currency later isn't a config change.
00:39:09.199 --> 00:39:17.599
It usually means actually re-architecting the ledger to currency to carry a currency field on every balance and transaction.
00:39:17.599 --> 00:39:23.760
That's a retrofit tax multi-market fintechs almost always pay.
00:39:23.760 --> 00:39:27.760
And it's visible in how different near banks have expanded internationally.
00:39:27.760 --> 00:39:34.639
The ones that launched single market fairs have generally taken years to bolt on multi-currency support property.
00:39:34.639 --> 00:39:39.920
But in contrast, there's the ones that built multi-currency support from day one.
00:39:39.920 --> 00:39:49.360
Likewise, and of course, like Reap, they could add corridors incrementally without touching the core ledger.
00:39:49.360 --> 00:39:57.039
So Reap skipped that expansion tax entirely, but not by choice.
00:39:57.039 --> 00:40:04.800
Parents is fairly direct that APAC's fragmentation didn't leave room to build that single market version first.
00:40:04.800 --> 00:40:25.599
What that means concretely is a ledger, the compliance model and the settlement logic were multi-currency and multi-jurisdiction from the first line of code, which is a fundamentally different and arguably harder engineering and compliance and business starting point than adding markets on later.
00:40:25.599 --> 00:40:33.199
Most companies don't choose that starting point voluntarily because it's slower to reach this first working version.
00:40:33.199 --> 00:40:39.360
But Reef's argument is that it's the only starting point that scales cleanly afterward.
00:40:39.360 --> 00:40:42.000
And I mean it worked for Reap, right?
00:40:42.000 --> 00:40:46.320
And huge congratulations, of course, on that.
00:40:46.320 --> 00:40:48.880
You had a really strong 2025.
00:40:48.880 --> 00:40:53.440
Um, how do you see this acquisition furthering your growth?
00:40:54.159 --> 00:40:56.239
Um, firstly, thank you for that.
00:40:56.239 --> 00:41:02.159
Uh we are all of us um at Reap are incredibly excited about it.
00:41:02.159 --> 00:41:08.480
So, from from my perspective, the acquisition actually doesn't change our strategy.
00:41:08.480 --> 00:41:12.719
The best part is that it accelerates it.
00:41:12.719 --> 00:41:22.400
And we've always believed the future of payments is built on combining stablecoin infrastructure with traditional financial rails, right?
00:41:22.400 --> 00:41:25.920
And and that's exactly what we have uh been building at Reap.
00:41:25.920 --> 00:41:38.960
What this acquisition does is gives us more of the infrastructure regulatory reach and distribution needed to execute that vision at a much grander and larger scale.
00:41:38.960 --> 00:41:43.920
As you mentioned, 2025 was already a breakout year for us.
00:41:43.920 --> 00:41:52.239
We tripled both revenue and payment volume, and today process more than 10% of the global cart link stablecoin volume.
00:41:52.239 --> 00:42:00.480
That momentum has given us confidence that the market is real and that we have very, very strong product market fit.
00:42:00.480 --> 00:42:06.400
Now, this acquisition simply gives us a much bigger platform to build from.
00:42:06.400 --> 00:42:09.760
So one of the biggest advantages is regulatory reach.
00:42:09.760 --> 00:42:14.239
Payward brings established licenses across the US and Europe, right?
00:42:14.239 --> 00:42:20.320
Opening markets and payment corridors that you know would have taken us years to build organically.
00:42:20.320 --> 00:42:22.239
The second is infrastructure.
00:42:22.239 --> 00:42:34.000
Access to much deeper liquidity pools, custody, and settlement capabilities means we can continue improving our products while scaling globally with much greater efficiency.
00:42:34.000 --> 00:42:37.440
The third is distribution.
00:42:37.440 --> 00:42:42.800
Hayward already serves nearly a thousand nine hundred B2B partners.
00:42:42.800 --> 00:42:51.679
Over time, those businesses become a natural audience for Reaps embedded card issuance and cross-border payment capabilities.
00:42:51.679 --> 00:42:56.880
From a product perspective, this is incredibly exciting.
00:42:56.880 --> 00:43:16.400
Instead of spending years, you know, building every market, right, getting every license and distributing and establishing distribution channels ourselves, we can focus on energy on what actually differentiates us, which is building products that make stable coin infrastructure useful for businesses.
00:43:16.400 --> 00:43:20.719
So I see this acquisition as a force multiplier.
00:43:20.719 --> 00:43:29.920
It gives us a skill, infra, and global reach to execute the strategy that we were already pursuing, but much faster.
00:43:30.480 --> 00:43:36.639
This is genuinely a smart acquisition for Kraken, and it's narrower than buy-in growth.
00:43:36.639 --> 00:43:42.559
Kraken already has liquidity, it has custody and settlement infrastructure at scale.
00:43:42.559 --> 00:43:45.280
That's the core of what it runs.
00:43:45.280 --> 00:43:51.280
What was missing was direct issuing rights outside of the US and EU.
00:43:51.280 --> 00:43:56.559
Visa principal membership is underwritten market by market over years.
00:43:56.559 --> 00:43:59.039
And Reap has already done the heavy lifting.
00:43:59.039 --> 00:44:02.079
I mean, it's done it in Hong Kong, Singapore, and Mexico.
00:44:02.079 --> 00:44:09.280
So some of the key Latan and APAC corridors that would be strategically important for Kraken.
00:44:09.280 --> 00:44:14.800
Now, secondly, let's look think about it from the Reap perspective.
00:44:14.800 --> 00:44:28.000
Everything in this episode, the collateral engine, the post-paid credit structure, the money market fund integration, they all depend upon collateral sitting somewhere and being managed well.
00:44:28.000 --> 00:44:38.639
Kraken's balance sheet and custody infrastructure gives Reap deeper liquidity to run that collateral engine at a scale that it couldn't reach independently.
00:44:38.639 --> 00:44:41.760
Plus, of course, it gives it a distribution channel.
00:44:41.760 --> 00:44:46.800
Harris calls it an accelerant rather than a strategy change.
00:44:46.800 --> 00:44:49.840
And on the numbers, that's an accurate read.
00:44:49.840 --> 00:44:56.639
Reap doesn't need Kraken to know what to build next, it needs Kraken's balance sheet to build it faster.
00:44:56.639 --> 00:45:01.440
Third, the mechanism underneath it all.
00:45:01.440 --> 00:45:08.159
The collateral engine is functionally a stablecoin collateralized credit facility, not a debit rail.
00:45:08.159 --> 00:45:20.000
That structural choice is what puts Reap's transactions on credit interchange schedules rather than the capped debit and prepaid schedules most cryptocards are stuck on.
00:45:20.000 --> 00:45:26.159
And it's why Reap holds full interchange directly rather than splitting it with a sponsor back.
00:45:26.159 --> 00:45:35.360
And of everything covered in this episode, the agentic payment architecture is probably the piece worth watching most closely.
00:45:35.360 --> 00:45:41.360
What Reap has actually built there is a delegated authorization model.
00:45:41.360 --> 00:45:45.199
A human science and spending policy once via pass key.
00:45:45.199 --> 00:45:50.639
That single authentication event gets referenced for every subsequent agent transaction.
00:45:50.639 --> 00:45:57.840
And the checkout challenge gets surpassed or suppressed because the liability shift already happened upstream.
00:45:57.840 --> 00:46:01.360
That's real infrastructure, not a roadmap.
00:46:01.360 --> 00:46:05.760
And Reap already sits on visa's agentic ready issue a list for APAC.
00:46:05.760 --> 00:46:09.840
But it's worth being precise about what's solved and what isn't.
00:46:09.840 --> 00:46:17.119
A scope token proves a transaction is valid within policy, not that it reflects actual intent.
00:46:17.119 --> 00:46:26.800
And Harris said directly that multi-agent delegation and agent scope narrow in correctly as it gets handed down to sub agents.
00:46:26.800 --> 00:46:28.639
Well, that's still being built.
00:46:28.639 --> 00:46:32.719
And that's something for the whole industry to really think about and reflect.
00:46:32.719 --> 00:46:34.320
And that's mint condition.
00:46:34.320 --> 00:46:36.159
Thanks for listening.