Visa Intelligent Commerce Explained: How AI Agents Will Actually Pay

Uncharted territory. Where road maps are useless. Conditions don't negotiate. The terrain doesn't wait. Where opportunity abounds, but so does death. The builders who understand what's coming aren't mapping it. They're already

We're talking about theoretically confusing things with AI agents and finance. You know, it's it's good to be a little perplexed every once in a while. Absolutely. I I was admiring the quality of the intro video, actually. I like that. Thought it was was nice. Thank you. Yeah. Thank you. Thank appreciate. There's a there's a little pun baked in with incline, the the climb toward agenda commerce. There's a little CLI hiding in the middle of incline. You know?

Little command line interface for our AI nerds or just computational nerds in general. But, Simon, we're we're really excited to dive in. I was looking on the the site, Sumvin. Is that how we're saying it right? Correct. Yeah. Yeah. And it says it says the last finance app you'll ever need, which I love. You know, we sometimes talk about the fragmentation of solutions. There are so many solutions that it's a problem. You know? It's a huge problem how many solutions there are. And this my little

my little simplicity loving heart was beating faster seeing this, the last finance app you'll ever need. Tell us a little bit. What are you up to over here? Absolutely. So first and foremost, great to be here today, guys. Thank you for having me. I think, you know, what we're really building out here is looking at kind of I've been in this space a long time. Right? Yeah. Realistically, in and around kind of crypto payments for the best part of a decade.

Obviously, I've worked with a number of the kind of leading teams. And I think, you know, across that time, what we've seen really is this kind of stagnation of, you know, product and experience. Right? I a lot of what we're seeing in the market is about fast following. Right? And what we were really looking at was the idea of saying, look. You know? If you wanted to really kind of look at how do you create next generation consumer experiences, AI really is the gateway to do that. Right?

And, you know, if you kind of go back here a little bit and we have a little bit of a history lesson, I think there were probably three kind of critical critical design features of the original web two or the Internet as it's more commonly known. Right? One was the lack of private keys, two was the lack of money supply, and three was the lack of identity. Right? And if you solve for those three things, you can actually start to create, you know, completely different consumer experiences.

And and what we really kind of, you know, as a business, what we look at and where we come at everything from is like that. We think a lot of data flows as they sit today are illogical. Right? And that's generally because the data exists, but there isn't necessarily a way to push that through a system in a way that's usable. Right? So the example I always give is, like, you know, if I wanna open a bank account, I already have the information I need to open a bank account because I know it. Right?

But I still have to go through this laborious process of going through an application where I have to put everything in at at the end of the process to be able to actually do that. So we were like, okay. Well, why don't we look at creating a system where we start with identity, which is a self sovereign identity, by the way? It's all completely controlled by the user. It's not a credential that we control. We just orchestrate it.

But, actually, then how you can downstream the effect of that how you can downstream the effect of that identity to actually be able to unlock various different things. Right? So the whole idea really being that, you know, you could move yourself from intent to execution, and you could put that all in one natural layer that consolidates things, enables you to make, you know, good decisions, but actually good decisions that you can execute on.

And I think there's a whole theme here around agents and identity that, you know, we'd love to explore with you guys a little bit more today. But I think, you know, not enough people really are talking about the the actual effect of personalization that you can have with an AI agent.

Yeah. Yeah. The personalization is interesting. And particularly from a consumer standpoint, I was just looking at kind of the narrative and the why behind this business. It seems like you have sort of a design ethos that bad design is all about when incentives are misaligned. Right? And the current incentive is, in some ways, to waste your time. It's like the financial stack wants you to use a lot of time and attention and to say that responsible use is is through monitoring. Right?

You should just have your eyes on all your accounts and all these transactions and all this kind of thing. What if there was which which really seems like a a pretty much table stakes standard. Right? Like, you should just keep your eyes on it. It's like, well, what if it's bad? And I was like, well, try to be reactive and bring it back.

It seems like with the kind of program programmability putting rules in place, there's a potential to take a leap up towards a higher order way to manage money, both from a sort of individual consumer standpoint, and I assume this would translate to at the institutional level too. Is that what you're up to here? Like, getting people to be able to be smarter and faster?

Like, what's the real problem that's being solved or the suite of problems that's being solved? Yeah. Absolutely. So I think I think there's I think there's a range. Right? And I think that that's the beautiful thing about AI is that you can actually put something together as a product that, you know, has multifaceted capability. Right? And then users should be able to compose and configure how they want their interaction to look. Right?

So my my account could look very different to yours based on what I'm dealing with it. But we really see it like a loop. Right? And the the key to this really is that first and foremost, you have to look at where friction exists and how do I remove friction. The biggest form of friction in any interaction is always data and identity. Right? So we solve for that day one by by design because we have your identity, and we also have data that you choose to share.

Now the downstream effect of that really, you know, you can look at it in whichever way you want. Right? We take a very virtuous view on this, which is that actually one of the best ways to help people make money is to save them money. Right? So we've actually, with our initial tooling, we partnered with Visa. We're doing a lot of work around agentic commerce.

A lot of agentic commerce really is designed to kinda give true price discovery, best execute, and also to start embedding things like rewards, loyalty, etcetera, which are those things that are a little bit too lazy to catch. Right? All We know we could save $3, but we can't be bothered to go and search for a code. Right? So we can kind of abstract that away. We can make LHN tick. But the second part then is, yeah, when you look at kind of You did. Actually, like, last moats in these things.

I think it's fair to say that the last moat probably people used to say is SaaS. Really, it's financial services. It's the repetitive task that you have to do to open all these different forms of accounts or interactions. It mean that once you've done it once, the last thing you're do is go and do it twice. Right? So what we look at is this idea of saying, okay. Well, let's use the price discoverability that you can get through an LLM where I could look up, you know, the best price in 3,000 places.

But there's also then kind of compressed down the kind of you wanna buy this, do wanna open account, do you wanna place this trade to a yes. Right? And just keep it very, very simple and abstract away the complexity. Because actually at that point, what you start to do is you start to shift consumer behavior towards the idea of, like, you now have the ability to manage these things with the power of kind of web scale intelligence in your pocket, which is something that we didn't have before.

Now I think the third part to that is that, you know, I genuinely believe that one of the biggest mistakes in product has been apps. Right? There's an overappification. If you go to your handset today I did this myself, actually, the day I counted. I had over a 100 apps on my phone. Right? I probably used five of them. Right? Yep. Now what that tells you is that actually, you know, the surface layers that I'm downloading are not things I need to interact with.

What that tells you though is that I should be able to pull some of those things and some of those capabilities into other applications that I do use. Right? And that's the beauty of when we look at an agent. The agent actually has the ability to go and fetch things for you. Right? Historically, you know, product matrixes are pretty static. They're based around kind of predictive behavior. And actually, what you can say here is, well, look. You know?

I've got money over here, and I wanna make some money on my savings. These guys don't do savings accounts. Let's go and find myself a savings rate that works for me. It doesn't necessarily mean you have to go out of domain to do it. You just have to be tooled in the right way. And that to me really is where you start to this unlock where you can start to kinda compress and compile and build out ultimately the next generation of super apps. Right? Because let's be honest here.

If we really want to see AI, you know, kind of get a product layer and we really want to see, you know, stablecoin proliferation or these other things, it's a natural marriage that this should be the point of infection. It's all about consolidating finance and then how you manage things more effectively and more appropriately.

Where where are you starting in terms of, like, what you at least want to surface to the human on the front end versus the things that you might be fetchable on the back end? Like, how do you how are you starting with the user like, what is the current user experience look and feel like versus what might be possible in in the future?

Yeah. Absolutely. So we take a kind of sequenced approach to this. Right? So the initial thing is that we're launching a series of smart commerce tools, right, which so you may have seen we're partnered with Visa, Intelligent Connect, and various other kind of platforms in that space. And one of the first things we're focusing on actually is moving checkout. Right? Because we think checkout's in the wrong place.

Like, to me, it's bizarre that actually when I know what I wanna buy, I know how I'm gonna pay, and I know where I live, that, yeah, the last thing I do is have to put all that information in and click buttons to pay. Right? So we're, looking at these price discovery tools, and we could say, look. Okay. Look up this thing for me. Find this item.

You know, run this errand for me, and and actually give me true kind of web scale search because it's a great use case because, you know, I can't go and look up 3,000 prices of a pair of shoes, but my AI agent can, and it can buy them for me if I tell it to. Right? So the whole idea there is, you know, we start off with this idea of saying to a user, look. You know, you've just bought, you know, a thousand dollars worth of things. And because we've used intelligent commerce here, we've saved you $200. Right?

And how have we done that? We've done that through a mixture of, you know, we've found best price discovery. But we've also found the fact that, you know, there was an introductory offer of some loyalty scheme that, you know, we signed you up to because you told us to, and you got $50 cash back. Yeah. All these different things that you can start to pull together. I think at that point, what you've done really is you've cemented the relationship with the customer that this is the best way to buy. Mhmm.

But in addition to that, you've then kinda said, well, I've saved you $50. What do you wanna do with the $50? Right? Because I think the idea that, yeah, you're gonna allow people just to kind of, you know, run wild on your savings account or your pension or whatever with AI agent is probably some way off, and it's gonna need a big generational shift and a lot of trust to be built.

However, in the interim, the idea that I can help you, you know, interact with products in a smarter way and give you an experience multiple times better because actually I'm compressing five or six different functions down to one. That's interesting.

I'm seeing you you kind of alluded to this, Simon, but I've seen this sigil, proof of personhood, errand, perpetual checkout, envoy, the money agent. These are sort of three personifications of different steps in the process that you need to make this real?

Correct. Yeah. So we we start from the baseline. So the identity is there. And I think the key to identity is to obviously answer this question between agent and bot. Right? But, also, agent allows personalization, right, which I think, you know, is a huge thing that we're missing in modern product. Right? You know, when I go onto American Express' website, it should be able to tell me whether I qualify for a card or not based on a signal that they did on arrival. Right?

It should be able to tell me which card I'd like to be approved for. So we look at identity as a kind of sole bound identity. Then we have, you know, our kind of mandates and our pint structure. So that really is the kind of plumbing behind how we find transactions and how a user enables a transaction. Now everything we do, we like to keep the human in the loop. Right? But we have a kind of autonomous spectrum. Right?

So you could say, for example, if I was gonna say pay my Netflix subscription, that can be a one time recurrent, hey. Pay my subscription mandate. Right? It's not high value. Right? But then you kinda go through the process based on the complexity, and you kinda want to come back so that person's always in control of what they're doing. If we look at errand as a kind of layer to that, so if you think identity, rules, governance, and then you've got errand, which is effectively a tool. Right?

Errand is a phenomenal tool if you want to do something like book travel. Right? High value discretion we spend with dynamic pricing. Because the key is I've already got your identity. So let's say you were looking at yeah. You're in New York. You want a flight to Chicago, and it says the flight's a thousand dollars on Friday. You could leave Aaron with a prompt that says, hey. Yeah. I wanna book this flight, but I don't wanna pay more than $700.

If you find it in the next six hours on Friday afternoon to Chicago, please book me this flight. Mhmm. Aaron can sit there. If it finds it for you, it will book it. Right? And we can be perpetually updating that scan. We can also do some interesting stuff around, like, goods and offers. So let's say, you know, you've seen something you really, really like. We really wanna buy, like, a a new MacBook is a good example. Right?

We all know that periodically, Apple products do go on sale, but they're also quite expensive. But fundamentally, most people are not that obsessive about having the absolute latest version of a MacBook. Right? So let's say the Mac's $2,000. You could put a heron in the back to say, if you can find this $1,500 or less, please buy it for me. And the minute it drops, we can buy it for you. So these are the kind of intelligent things.

And this actually leads into what we call the kind of merchant of future category because when you think about it, we have the agents that have the orders. And as the merchants come online and they have their merchant agents, theoretically, you can start to create a kind of peer to peer network between consumer and merchant, which is where we see the unlocking of all this. It's about resetting that relationship between consumer and merchant.

What that does, of course, in itself is mean that, you know, you can go to a retailer and say, look. If you were to sell at this price, I could sell a 100 of these for you tomorrow. That's real. It's not just intelligence. It's not just guesswork. It's actual actionable orders. Right?

And then when we go down, obviously, we talked about perpetual checkout, and then we have, you know, a few other bits as well where we've done things like we've extended x four zero two so we can index effectively to merchant agents as well, which is something that I don't think anybody actually does commercially. So we can actually, you know, we can actually look up we look up about a million websites every day. We look at their UPC servers. We index them, and we confirm who they are.

And that's actually quite big because when people are using crawlers, you need to be able to verify that, you know, this is actual genuine Best Buy. It's not, you know, someone pretending to be Best Buy who sold your television, you're not gonna get.

Simon, you mentioned having been in the space for a while, operated with some big players, have had a pretty significant acquisition at at a time. Like, you've seen a lot. Is there something with the rise of agentic finance or commerce in the last couple of years that has really changed your mind? It's like it's it's like, wow. I didn't know this was possible. Like, you're really seeing something

new and different here. I think for me, I mean, first and foremost, agents and agentic tooling is getting better. Right? It's still very early. I think there was a period of time where, yeah, I would look at some of the LLMs in the market from sort of frontier labs and be like, I'm not sure I really wanna use this. Right? And that was because they were writing code. But for me, it's the traversions of really two things. I think it's AI and the evolution of AI plus stablecoins becoming good dollars. Right?

Because, actually, the Internet needs a money supply, and a stable is the most logical money supply. Right? So to me, that's a kind of perfect marriage because, you know, come January next year, it's like America where you can say, hey. USDC, USDT, you know, whichever other stablecoin that's chartered is is appropriate is a good dollar. Right? Suddenly, you know, I can start to then say to my agent, well, you can pay like this. Right? Whereas historically, the issue you always got was settlement. Right?

Ironically, now settlement should be compressed into one function, and that's where the whole machine to machine infrastructure becomes interesting. I think before that, it was always a little bit too high friction. And I think let's be honest. I mean, I've been in this space for, you know, ten years, and I'd say that, you know, there are probably a few companies out there, people like Coinbase, who've done a very good job of obstructing friction away from users.

But on the whole, you know, most platforms in Earth space really just added friction or complexity. I think now we're seeing this huge desimplification of things, which should lead to a very clear path forward for how kind of digital money really has a distribution line.

I wanted to ask you about intelligent commerce with Visa. Sounds like this is something that you are you're following very closely and also building, sounds like, directly on top of. I I found this linked to one of your blogs. Maybe walk us through a little bit of, like, what is Intelligent Commerce? How are you using it? What are some of these key benefits specifically that you find really helpful right now as you as you work with Visa? Absolutely. So I think, firstly, it's the simplification of standardization.

Right? I think one of the big things you see in an emergent kind of, you know, market is that everybody writes their own standard. And the problem with that is that you then end up with 18 standards, and then someone says, actually, those standards didn't work, so we're gonna write another standard. And then you end up with 19 standards. Right?

I I think I think where Visa's great is really, you know, first and foremost, the pull factor of Visa to actually, you know, kind of help platform the world away from the kind of traditional world towards the digital world and saying to some of those more incumbent players, these more traditional conservative players, hey. This is alright. Yeah. This works. Right? Is a huge thing. Right? Because ultimately, they are the validator for payments. Right?

There is no one else really you know, even Google or Cloudflare couldn't legitimately put their hands up and say we really understand how payments work in the real world. I think that they give you a very clear kind of underlying set of tooling that you can work on. They also give you an API that you can build over the top of and the which you can obviously build your own function like we did with things like Arund. Right?

But I think also the other thing that they do is they bring this kind of coalition together where you have people like Cloudflare. You have all these different people behind this who say, hey. Actually, these APIs these agents are okay. You should let them through your firewalls, etcetera. It is a big thing. And then I think that the last thing they do really is they give you that kind of pull to be able to go back to the merchant and say to the merchant, hey, mister merchant. You know, you should accept this.

There are news tokens that you can accept. There is new data. There's new insight you can get. And I think going back to my previous point, think I've touched on this yet. I think a lot of people in the agentic world, they talk about agents as a benefit to the consumer. Right? And absolutely, they should be. Right? But there's also two sides to that coin. You have to think about, okay. If I'm gonna spend money, I'm gonna do things. What's the real benefit to the merchant or the end provider? Right?

What do they get that they can't get now? What data can they get? What actual insights can they get? What personalization can they get? How do they not lose the relationship with the customer? So we build out a whole bunch of things based on the identity platform that gives us this ability to share a raft of data with an underlying, you know, merchant if the customer wants us to do that. So actually, you know, if you look at it now, people are like, hey. You know, have you got cookies enabled? Have you logged in?

Well, if the answer is no, I mean, who cares? Right? But secondary to that is that we can actually give you so much more data and create that personalization. We go to the merchant and we can say, you know, this is something you don't get now, and this is something you get in the agentic model. This is why you should take this because actually you you can get actionable order flow. You can get better insight to your customer.

You can understand more about where your customer comes from as opposed to just a sex location when they last visited your website, which is what you get as a cookie. And it's really interesting. We're actually with a fairly major airline last week who specifically have it in their terms and conditions that they will ban anybody from using an AI agent to buy a ticket. Right? And we we talked to him about this.

It was a fairly, yeah, traditional response, which was like, our concern is that people will buy tickets they didn't know they were buying. And I understand that. Right? So we talked him through what we were doing, and the guy turned around to the end. He said, look. Can you tell me one thing? Could you tell me which passengers are booking tickets on my airline that are frequent flyers on American Airlines? And I said, the customer tells me that, like, I can show that information, and, yes, I can. He said, okay.

Great. I'll let you I'll let you use an AI agent to book a flight. You can tell me that. And and I said to him, okay. And as to me, it's not interesting. What specifically about it? He said, well, we're doing a status match.

So at the moment, if you've got a flight status match on American, we can you know, if you can tell us the other airline, which I'm not gonna, you know, I'm not gonna name who they were, if you can tell us which fires have, you know, sapphire or platinum or whatever on American Airlines, we're gonna max their status automatically so that they get all these benefits on their first flight out of this. And I was like, okay.

That is a great example of something that works really well possibly for anyone but American Airlines. Right? Because as a customer, you're gonna get benefits you weren't expecting with your booking. But, obviously, as an airline, you're also getting this kind of qualified, hey. This person flies a lot. Yeah. This person is someone that you could win. And you think something that's a virtuous circle. We're just deploying a bit of data and a little bit of actual insight. Yeah.

Well, yeah, I I know my frequent flyer status, but I don't have a way of telling another airline what my frequent flyer status is on a different airline.

So Yeah. It makes me think what we're throwing around agentic commerce, but it's intelligent commerce broadly. There's more data, more insight, better incentives. That's where we're headed, from that example.

I just think my point would be, yeah, if you're just gonna wrap for your people up now, why would you run shift? Right? We all know we all remember a product. The first time we used it, we thought, okay. I'm not getting back, right, to what I used to do. And that's really what we need to be thinking as well.

Let's do this. We would love to Simon, when you have, like I guess the next step for us would be, obviously, if if one of these big airlines become a client customer partner or whatever, like, we'd love to have you back on the show. Or or really just, like, any any of these use cases, the Aaron's product specifically is, like, very, very compelling. I also, one, love the name of that. I think that's just so creative and perfect. But, yeah, let's would love to have you back.

And then when when we do, maybe bring you can feel free to invite a partner, client, user, and we can kinda walk through some of these specific use cases. That'd be fun. But thanks for bringing us into your world a little bit,

and we're excited to keep following along. Thank you. Thanks for having me, Goose. Thanks,

man. Appreciate you.