Rishabh Jain:

All right, welcome everyone to another episode of The Intelligent Marketer. Today we're excited to have Rubail Birwadker from Visa who leads their growth products and strategic partnerships. I had the benefit of meeting Rubail, actually, at a dinner that they hosted recently where they invited a bunch of startups in the agenticic commerce space and really that's kind of where I want to start, Rubail. I was curious to just start off with you giving us a little bit of perspective on how you think about the role that you and Visa want to play as this new world of agenticic commerce starts to unfold versus, I think, what a lot of people might expect from larger players, which is more of a wait-and-see approach. You guys seem to be coming out of the gates really pushing on the innovation, so just help us understand how you think about the role you want to play.

Rubail Birwadker:

Yeah. Hey, thanks Rishabh. Thanks for having me on. It's an excellent question and I just think it goes to the DNA of how we think about our business at Visa. I talk about this quite often, which is if you have a Visa card in your wallet, and I hope you do, and I don't know where you're based, if you're based in San Francisco or New York and you can fly to 190 countries around the world from one of the airports. You could whip out your Visa card and it works almost exactly the same no matter where you are — not just in a physical environment, but an eCommerce environment, in a app-based environment, in a mobile browser environment. That has taken an enormous amount of work for us to make sure that our network remains compatible with wherever the future technologies are headed. So the one thing that we wake up every day and go to bed is to figure out how do you make a network with 10–15,000 banks, near 150 to 200 million merchants, four and a half billion credentials around the world, and how do you make it compatible with things like agentic commerce, with things like stablecoins, with things like wherever the next technology is headed.

So in some ways, we actually can't afford to take a wait-and-see approach. We almost have to define how we believe the compatibility would work. Ultimately, we're a network, and the best part about networks is we can design standards that ensures that things work almost nearly the same almost everywhere, which again, then translates into a deeper level of trust, a deeper level of consumer experience, which is consistent, where consumers are like, I trust using my Visa card because my money is safe. We want to make sure that that happens in Tel Aviv and it happens in Mumbai and it happens in San Francisco and it happens in Bogota, and we want to make sure that that consistent theme percolates no matter where you are. So it's an innovation imperative for us to stay ahead.

Rishabh Jain:

I think one of the things that I was really quite impressed by is, as you are defining standards and approaches in agentic commerce, the inclusion of the startup community was very strong as well. And again, I'm curious what motivates that approach to defining the standards. Again, another way that you could imagine someone from the outside looking in would think about this is, Hey, you have all these partnerships with these banks, why not start there? Or maybe it's like, Hey, we actually have to do both. Yeah, just help us understand how you think about the way that you do it.

Rubail Birwadker:

Yeah, thanks for that question and thank you for the compliment. I think a lot of that deserves with our executive leaders. Visa is an enormously tech-forward company, at least in my opinion, and we are very lucky in many, many ways because we are based here out of Mission Rock, and within, whatever, 25-mile radius, you can throw a stone and hit almost every important technology company around the world that we work with or want to work with. The geographic co-location is an enormously helpful thing. Second, we've always been more tech-forward with our partners, especially in Silicon Valley. I think we were early investors back in I think 2010, 2011 in Square. We were early investors in Stripe, in Marqueta. We've largely stayed very, very close to the community and, especially in a world like agentic commerce, we know what we don't know, which is: We are not smart.

All the bright people that move from all around the world and raise a lot of money to go figure out very, very hard problems. So it would really be a waste of the privilege that we have to get to do what we get to do without being super plugged into the community, something that I've deeply been engaged to over the past decade or so. But to answer the second part of your question, it's both. Look, I think almost anything that we introduce new in the payments innovation, payments ecosystem, again, I've come back to the compatibility piece: It has to work for our banks, it has to work for our hundreds of merchants, it has to work for several billion credentials. Without working with the ecosystem, we don't want to do things subscale, they have to work at scale, and that's where, really ,the power of the Visa brand really shines.

Mike Duboe:

Rubail, I think it might be helpful to take a step back and just get your definition of agentic commerce. I ask it at that altitude because we've had different guests on the show from a marketing background, from eComm merchant background, from people talking about answer-engine optimization, but I think you're at a very interesting vantage point to define the subject. How do you all define it?

Rubail Birwadker:

It's a hard question, Mike, and I hope whatever definition I'm about to give you — and I will give you one, believe me — I hope it doesn't stick because ultimately we want things to work regardless of how and where the industry evolves and these channels evolve. But if I was to take a stab at saying what do I believe truly that's agentic commerce in my mind, this is how I would define it. I think commerce is getting enormously intelligent over a period of time, and as commerce is getting more and more intelligent, it's just like there's a lot of things that are moving toward the edge of where consumer is actually taking an action or an intent to buy. Discovery's getting smarter, data personalization is getting better, there's enriched data payloads to make sure there's better identification of fraud and friendly fraud and all kinds of things that could go bad.

So the cleanest way I define where I believe where agentic commerce starts is when you are in an environment that is not truly a merchant environment, but you're asking a piece of software to go do something for you on your behalf at a merchant. To me, that's the clearest defining line of where agentic commerce becomes very, very clear. Now, I'm not talking about delegated purchases, I'm just, for example, typically I am at Macys.com or I'm in a Macy's store, I'm on AMC.com or I'm in an AMC theater. I know clearly, I am in a merchant environment, I'm using my card or a different payment method to buy something directly from the merchant. I'm clicking on a buy button or I'm tapping or swiping or whatever I'm doing with my card or I'm using my phone to pay for it, but I know I'm directly interacting with the merchant.

Now, in the new world, that's not so simple because what's happening is you have all these new intermediaries that are adding an enormous amount of value that sit in the middle. AI platforms are a really, really good example of that where they're helping you curate, contextualize, compare, personalize, and then when you're ready to change that into an intent to buy, they're going to take your identity, your credential, your data payload, go to a merchant, and make a purchase on your behalf. That to us ... And it could be immediately, it could say ... An agent would show up four types of red socks from four different places. You're like, Well, I like the left one, buy it for me. It takes your ID, an authenticated credential, it goes and buys and says, I just bought this for you. Or it could be, When it goes below $30, buy it for me, and it does it in two days or six days or eight days or whatever else it is. But in both cases, it is taking your intent and it is performing and automating a thing that you would ordinarily do coming from a context. So for me, I know that's a mouthful for a definition, but that to me is a very clear line of where agentic commerce starts to really shine and really starts to come into its own.

Mike Duboe:

Okay, that's awesome. Maybe we can parse out different stages of the funnel there. I mean, I think of this as a funnel, and you talked about clearly, right now, the discovery process — how we actually discover products — that's in the process of being reinvented through AI platforms. You mentioned that agentic commerce does not necessarily require an agent to go ahead and handle a transaction without the consumer being in the loop there, but there's a bunch that happens in between that is in the process of being transformed by agents. Maybe as you think about existing eComm platforms and what they're doing well in this and maybe what gaps exist in the existing commerce stack, how would you answer that?

Rubail Birwadker:

It really depends. The thing is, existing eCommerce platforms already are bifurcated in so many different categories. There is the nike.com, which is clearly a first-party environment with first-party logistics, for example — I'm picking on nike.com, but it could be any large brand and what they actually do. Then there is first-party and third-party worlds of, say for example, amazon.com, relatively different in the way they interact with the users and how their brand is perceived and what kind of logistics do they really provide. Then there is things like walmart.com where there is first party and then there is things like endless aisle. Then there is things which are less to do with physical goods and services, but more on adding a layer of value, on aggregation, and making your life a lot easier. Things like booking.com. And then there is the long tail of, say for example, a direct to consumer, like a typical Shopify website, where you just discovered something that you liked.

And then there is a whole bunch of other surface areas where you see a lot of eCommerce show up. Instagram's a really good example of that, but it's not really confined to Instagram. Almost every other large social platform of some kind have some form of shopping enabled. So they all have a little bit of a different way they operate, the way their payment stack is, and the things in which they relate to their end consumers are. So it really depends, Mike, which way we want to talk about it, but they're all going to take a different approach. My belief — again, this is ... I wouldn't assign this to anyone, but just my personal belief — is, who knows, I don't have a crystal ball, but they're all going to have to adapt a little bit as to where this is headed, and they all will become very intelligent.

I think a lot of commerce, I personally believe a lot of eCommerce is going to become very intelligent, and yes, agents are going to drive a lot of it. But a lot of these examples that I gave you, many of them may become agents themselves in order to provide a personalized experience where, as a consumer, say, instead of going to an agent context window and asking for the best pair of socks to buy, if you really, really have that affiliation and association with nike.com and you can trust it to give you the best personalized things and you already have a long-term customer loyalty with them, maybe the chances are you'll always just go to nike.com and sometimes you're not nearly as cost-conscious. It really depends. I think we're going to see many, many flavors of what this really looks like.

Mike Duboe:

So maybe if we just take an example — and you're right to call out that these are ... the term eComm platforms is not very helpful. So maybe what's in my mind, let's spend a minute on Shopify. They've announced they're doing some stuff with OpenAI. I think their merchants tend to be very forward-thinking or at least quick to experiment. What's your observation on what they're doing so far here?

Rubail Birwadker:

They're an awesome team. I think they're very smart and I think they're going to try a lot of things to make sure that their merchants have the single best experience for selling their goods and services. I mean, that's like my top-line observation about everything I've seen the Shopify team do. I think they're incredibly bright, they're very smart, they're very thoughtful. We spend time with them and I think they see a lot of power in what cards and networks can do and add, because ultimately it's a very, very consistent experience. I think Shop Pay I think is a great, great buy button and it's largely a card-based product and it's deeply integrated into our infrastructure and ecosystem. So look, again, I don't know exactly what they're thinking and where they're headed. All we do know is there is an enormous amount of work that they will continue to do, which I think will be very, very compatible with the Visas and the other card networks of the world.

Rishabh Jain:

Yeah, I think, Rubail, one of the other things you were mentioning, tying this and one of your earlier comments on your definition of agentic commerce together, this idea that you're going to trust a piece of software to make a buying decision for you. I know one of the things that you guys have published is this view on a token-based approach that Visa is taking to enable that. So that way you can certify that, Hey, this is actually me authorizing, that this is a payment that I want to make. How do you think about that approach versus other potential approaches that you considered when you were thinking about what actually needs to be true? It's very clear that one of the main things you were optimizing for is trust. Trust and safety, I guess, is a way that I would characterize it, or at least that's how it came across to me as I read it, is just like, Hey, I want to make sure that, no matter what, you can trust that this is actually this person who's authorizing this. How did you think about the trade-offs of the different approaches? Why did you land on that one? What were you optimizing for? And then, what do you think merchants should be thinking about in that context?

Rubail Birwadker:

Actually, we do debate, we still debate, we think about this a lot, and I know you know this because you see this a lot in the industry. There are several different ways of how potential commerce could be enabled in an agentic commerce environment. But I think you hit on the right points and I can explain a little bit of our thinking, which is it's trust and safety. I think that's absolutely right. It's also scale, and native, and a native experience, so you can stay within the app and still make the transactions happen without ever leaving the environment. So those are the three things, which was trust and safety, scale, and the nativity of an experience. Let me address the first two and I can explain a little bit of our thinking. So, first of all, right, look, we don't know what the new vectors of fraud are going to show up in a world where you're entrusting a agent, an LLM, a piece of software, to go make a purchase on your behalf, even though it could be a relatively straightforward purchase, it could be a delegated purchase, whatever it might be. The reason why I say that is because, the last 30-something years, the eCommerce industry has worked very, very hard to keep fraud at bay, keep bots at bay.

Identity and verification has always been a complicated process and you can see this around the world, literally: the 3DS standards, step-up standards, localization standards, depending on which market, when you look at them. When we think about payments, we think global. We obviously want to make sure that these things work in the United States is an important, very, very, one of our most important, most strategic markets, but it needs to work everywhere else. So just going back to trust and safety. So when we got together, the first thing we thought about was, how do we make sure that whatever we design has as little surface area for things to go wrong. That aligns well with the funnel in the way we actually designed it. So first of all, in the way we've designed it is, we kind of want a authenticated token. So token, because we don't want your PAN and your PII floating around sitting with a agent, because, well, that's not good.

Suddenly now you've got someone selling a bunch of PANs on the dark web. You don't want that. So tokenization gives you the first level of protection. Authenticated — and when it's authenticated, it's not dissimilar to any of the pays, Google Pay, Apple Pay when you use — it binds a token, which is a representation of a card, to a device and an app. Now, there is some friction there. For the first time you add a card, there's always, there's a step up or there's something in the background, we know like, okay, this consumer is good or bad, where we do a call out to the bank to make sure that ... Because that's a very, very good source of trust. So once you actually get the provisioning in, after that, all transactions that happen typically have very, very non-material fraud because you've already gone through the process of really identifying: This is Rishabh on his specific big-bank card on ChatGPT on his iPhone 10.

Anything else that happens is not going to be allowed on the token. So already you reduced the vector of fraud by 90%. Then, as in when you have more human-in-the-loop transactions, you'll start to — what we call the CDCVM or a biometric authentication at the time of purchase, if that's what you really choose to do — that further reduces the amount of fraud that you could get. So you further narrowed down the focus on the vector and how that worked. Then the third piece is, we've actually designed a bunch of data payloads in what we call payment instructions and signals as part of our product design where, before we allow a card to be used by an agent, while an agent may have a card, it can't just go and take that card and buy something at a merchant, it needs a underlying cryptogram, which makes it a very safe and secure transaction to make that purchase happen.

And we don't issue that cryptogram until we get a data payload confirmation that what an agent is trying to buy is aligned with what the consumer asked in its natural language prompt or whatever the prompt was. So we have a little bit of a matching process that happens. Then we issue the cryptogram, the agent takes a cryptogram and then goes and does a purchase on behalf of you. Now we further reduce the vector of fraud. Truly, without designing something like this, we have no idea how and what the types of frauds we'll see, and look, it remains to be seen. We are very, very early days. Ask me this question in a year when we have hopefully many, many million transactions in here, and then we can look at it and say, oh, maybe we overcorrected, maybe we undercorrected. But when it comes to fraud and things like that, it's better to overcorrect than undercorrect because then that goes to the trust piece.

As a user, you're like, Who is Joe Pesci? When you ask ChatGPT, great, it could get the birthday wrong or whatever it is, but you don't want to make sure ... You want to make sure that that level of precision is not lost when it comes to your money. So we are deeply, deeply invested in making sure that your money is safe. We want to see, we learn from the data, we want to see what these transactions look like, and after a period of time, then we can adjust what and how and what kind of authentication and other things that we provide. So that goes to the point I was making around fraud and trust. And then the second piece is scale. Most of these LLMs that you talk to or agents that you talk to or any big tech company that you talk to, if you look at OpenAI or some of the others, they'll tell you not only a majority, a large, large percentage, and I don't know exactly where the numbers are, but 70–80% of the users are outside the United States.

Now, that's good, but you want to make something that actually works outside the US. How would something work in Europe with their 3DS requirements? Now, 3DS is a very specific European PSD2-related ... There's a lot of regulation and there are data requirements that come in. People gotta get a chance to step up at the merchant level. What would that look like and how would we design that? Which is one of the reasons why scale matters. Or if you want to try this in Australia, they have their local routing requirements. Try this in Canada, there is Interac. You try this in Mexico where ... Just every market is different, and it's very hard for payment methods that are designed for a relatively narrow use case to actually scale outside of the use case, across multiple geographies, across things like commercial and what we call the large market or small business or consumer or whatever the type of cards might be. So one of the reasons why we like the technology that we've designed is we have and we understand how to scale it to the globe versus stay restricted to a geography or a use case. Those are the considerations we spend a lot of time thinking about and we still do. We don't have everything done. We have a lot of wood to chop, but that's why we were spending a lot of time.

Rishabh Jain:

That makes a bunch of sense. One of the things that I think about, again, as a potential partner startup in the ecosystem, when I observe a lot of the work that you all are doing, there are two things that emerge from this — and you tell me how you all think about this — but there's one side of it which is, okay, as a merchant, I can now trust this payment because of this work that Visa is doing so I can actually accept that this is going to happen and I don't have to worry about things like chargebacks and all of the other stress that you have to worry about if you can't actually trust that the payment is going to be real. On the other side, it basically causes a big question to be asked, which is, what is actually the system that the consumer is using to make the transaction on their behalf?

So is it actually ChatGPT or is it some other new startup that is a shopping agent for you? There's lots of ways now that you can imagine software being built that allows for this type of transaction to then happen with the end merchant. When you think about the parts of the ecosystem that you want to push on, are you thinking, Hey, I want to identify all the interesting ways that a consumer may actually want to shop with the merchant and make sure the merchant is supported? Or are you spending time more on the merchant side and saying, Hey merchant, no matter who comes to you, we want to make sure that you're ready to go? I guess, because it is ultimately a network and there are two sides, how do you think about where to spend the time, where to push on the innovation — or learn from the innovation — and then continue to make progress?

Rubail Birwadker:

Yeah, I think you're pointing to ... It's almost like you're in my meetings sometimes, Rishabh, because you're reading my mind as to where we spend a lot of time. There's actually one more vector. I actually address all three of them. So for the past 20 minutes or so, we've been deeply focused around how would a agent make a transaction happen? But then there are two other sides to it. What happens on the bank side and what happens on the merchant side? Now, if you go talk to banks — big, small, large, anywhere in the world — they will consistently tell you they're very excited about agentic commerce, but they worry. What kind of, say for example, call center traffic are they going to see? Are consumers going to call and say, I went on certain AI LLM, I asked them to buy something, this is not what had happened. So they have to service all these customers.

And servicing customers, it's expensive. That's one. Second is, as a bank, you have certain standards, how you might deal with a certain dispute. Now a consumer might say, I never said "buy that." I said something else which sounded like "buy that." The intent is not ... Today it's easy: I click on a button. That's easy to capture. But things can get lost in translation when new channels like agentic emerge. So we are doing a lot of work on our dispute, chargeback, fraud rules, to make sure that they're clear, it's known, it's more actionable, and hopefully we are catching a lot of these disputes before they actually even become disputes. On the merchant side, I think there is a similar issue. You talk to merchants and they're very excited because now a good some percentage of their traffic is showing up because of AI agents and bots and other companies that are building interesting things and they're showing up.

And as a merchant you have to make certain choices, which is how do I expose my inventory pages? How do I expose my pricing details? How do I expose my order details? Because if a bot shows up with an intent to buy, that's good for me, but maybe a bot shows up with a different intent and maybe that might not be so good for me. The merchants have looked down this road already and it's like a lot of smart reimagining of the internet already happening at the CDN level where there are these protocols being used by various companies to look at what's a good bot, what's a bad bot, what are they allowed to access or not? Our job here is to get plugged into that infrastructure so that they know that when a agent is showing up with an authenticated, identified Visa credential, it has some form of intent to buy and a merchant is welcoming it with open arms and saying, Oh great. Here, this is what might be an easy path for you.

Again, I shouldn't say this to you all, you already know this, which is the merchant environment, merchant websites, are not designed for AI. They're designed for human beings, just like font color, sizes, buttons, interstitials, an agent doesn't care about any of that. So there has to be almost like a fork in architecture to service these agents that show up. And how would it make it very, very easy for the agent to actually get through in a rapid amount of time in order to make a sale happen? So we're spending a lot of time on this. Most of our time right now on the merchant side is spent around making sure that merchants see a bot showing up with the authenticated Visa credential as a positive thing for them to let it through at the CDN level. Sometimes it's not even a payments problem, it becomes a payments problem when it hits a processor, but this happens so much more upstream that we have to figure out how to solve for that.

Mike Duboe:

I want to spend a little bit more time on one of the points you just made, which we are seeing interesting early startups working on this problem too, is the notion that websites are designed for humans, not agents. At a high level, that's conceptually interesting, but then when you get down to specifics, how do you actually optimize a site to convert agentic traffic? Yeah, I guess it's not immediately clear what one does there if they're a merchant. What are you seeing there, and maybe can you say more on that point on not only what's Visa's role there, but actually what actions are you seeing merchants take to actually inform that conversion?

Rubail Birwadker:

I think Shopify is a good example. I think they've made their MCP servers available, they made it easy for agents to navigate a Shopify website when they actually want to buy or peruse or look at the order details. I think that's as tech-forward as any other company you could look at from a merchant perspective. And I just think they're going to see more of that. I just think it's going to be a redefining, refactoring of the architecture to accommodate for agents to show up. I think that's the bottom line. Now, that's easier said than done because that means it's a lot of work and there's not enough expertise in the industry and I think there's going to be an enormous amount of — there already are — and there's going to be more enormous amount of companies that are going to be built to help merchants navigate the new reality of ... It's almost like when you had a physical store and then people are like, There's this thing called the internet.

They're like, Wow, so you need to have a website. And then there's all these cottage industry of companies that help companies make a website, so human beings can go online and peruse and buy things and look at things. And I think, slightly at a different, faster, cleaner scale, we're going to see a lot of that around, I think, around AI. The thing that I have no idea — and I'm not ... I'm a payments guy, I have no idea how ... I don't know how to sell anything, and merchants are incredible that they figure out how to make and sell things — is I don't know how brand is going to evolve on the merchant side. I look at it from the outside and I'm just as curious as anybody else, which is what would brand look like for a merchant, for a large brand to a small brand? And that's a complicated question.

Does having your own agent with its own brand associated with the user is the way to go for a merchant? What would high-end brands do? Or really, really expensive brands do? I have no idea what they're thinking. I'm sure they're thinking of something. And what would a small, small merchant that advertises on Instagram to get eyeballs to get ... What would they do? What would that look like? I actually really truly think those are fundamental questions that really, really need to get answered. And we've started to see some early signs of where that would go, but I think we're still very, very early days. From Visa's perspective, I think the answer is very principled, which is, we want to make sure that anything we build is accessible to the largest, most important, most enterprise-wide companies on the merchant side to the smallest individual website selling widgets of some kind. And as long as — when a Visa credential shows up in an eCommerce environment, in an agentic environment — the processing, the dispute, the chargeback, the guarantee of our brand works exactly the same way. As long as we can do that, I feel like we've done our job and the rest is we just have to see how it plays out.

Mike Duboe:

All right. So you made a bunch of great points there. We should come back to the brand point later, but I want to spend a minute on the assertion that there will be a lot of interesting companies innovating around this new theme, which we agree with as well. I think one of the questions we ask ourselves actually as investors is, there's a bunch of point solutions, let's say around catalog feeds, like helping merchants API-ify or create MCP feeds for their catalogs is clearly something that is necessary in this world and there's a number of point solutions doing that. Maybe that's just a wedge. And then there's some companies that are just trying to go and be like a next-gen Shopify, just handle it all. There's an interesting question there on point solutions versus a true end-to-end new platform, if you will. Do you have a point of view on, if you're an entrepreneur going and working on this, what is the more interesting path? Is the latter path actually even possible right now?

Rubail Birwadker:

I think it is. I think, again, I don't even think we're at the bottom of the first innings on this one, Mike. I also — and it depends who you ask, right? I typically like things that are global because I think the one thing we know for sure that technology is doing — AI especially, but almost all the technologies over the past couple of decades — what we've seen with cloud and mobile computing is they really are democratizing, democratizing talent and democratizing people to sell across borders, sell and do business across borders. Now, obviously there's a regulatory overlay to that, so let's just put that to the side for a second. But if you truly believe what technology is capable of doing, it doesn't matter where you are in the world, you should be able to sell your goods, services, intellectual thinking to anyone else around the world, which is a great opportunity for Americans.

So it's a great opportunity for non-Americans to do business with each other. It hopefully brings down technological barriers to do that. And if we actually truly believe that, then I think the TAM is so enormous. We have eight-plus billion people in the world — still growing, as far as last time I checked. So I don't know, I actually think there's not just room for one Shopify. I think there's room for multiple Shopifys because we still don't have ... I mean, one of the reasons why we like — just to tie this back to the stablecoin piece — one of the reasons why we like stablecoins as an infrastructural play is because they are filling a gap of product-market fit in many places where cards don't fill a gap for product-market fit, which is in many markets with high currency volatility, like less digital infrastructure, consumers are like, I want to hold money in stablecoins.

I like my money being safe and tied to something which is a lot more stable in a market. And guess what? In most of the countries where we're seeing a lot of stablecoin adoption in emerging markets, typically we don't really have a lot of people domestically using cards. Still, there are multiple billion people in the world that are still very, very early in their digital journey. So I know that's a little bit of a cop-out answer, Mike, but I actually think ... I don't think the playbook's been written yet between point solutions or a scale solution like a broad-based company like a Shopify.

Rishabh Jain:

I'd love to double-click on this and, like Mike, I also want to go back to the brand question, but this international thing has me extremely curious what you are seeing in terms of early adoption on agentic shoppers. And you brought up stablecoins as a good example; in the eComm world, the prototypical example is live shopping. People will say live shopping is very popular in Southeast Asia or Asia and I mean, it kind of picked up in the US but really only Whatnot, it's not really a standard way or a very popular way that we shop here in the US. And so I guess when you look at the spectrum of different types of agentic shopping that you are supporting right now, where are you seeing early signals of the consumer mechanism? Or what are the types of apps that are actually showing real transaction volume? And more interestingly, how are you seeing differences across different geographies?

Rubail Birwadker:

It's a very detailed question, a good one. I'm not sure how much I can share, by the way, very, very early. So on transactions, I just think it's very, very early, very, very early. But purely from a adoption interest UX perspective, look, I think we are seeing a lot of interesting use cases and really smart thinking coming out of the more MEGA wallets that are based in Asia and things that have never really took off in the United States, like the big, large Asian wallets that have hundreds of millions if not billions of customers on their platform and they do all kinds of services on the super app. We've never saw a replication of that in the United States, but those kind of companies have always been on the front end of innovating, and we are starting to see them ask some really, really good questions, look and design some really good experiences, which are even on the more further of the front edge of the agentic commerce because they already have a captive audience of hundreds of millions if not over a billion users.

And how do you make sure that you give that captive audience what they need in a native environment without actually leaving that environment? Because, in many cases, users live a good portion of their lives on those apps. So we're seeing some really, really cool things in that and we are really enjoying doing some of the work with clients like that. Now, how much of that is really going to be instrumental to what they already do? What is the agentic environment going to look like? How is that going to interact interoperably with the rest of the world remains to be seen. We still have ... I mean, nothing that I'm seeing is not already public, but you already have very, very successful companies that have captive audiences that are now thinking of agentic in ... How would you think about agentic commerce in an environment where it's already in a logged-in environment and you get a full end-to-end service?

I'm not trying to be cryptic, it's mostly around just we are so early in the ideation stage and we're seeing all kinds of very, very interesting things, what these companies are doing. So I get super excited. The good news also is it's the number of banking partners in many of these markets are not as many as banking partners that are in the United States. So ultimately in order to scale in a country, you need to get six, seven, eight banking partners on on the card side and the thing works for almost everyone. Less so the case in the United States because we just have a much, much longer tail of banks that we need to make sure that they can support. So just every market's a little bit different, but I definitely think markets in Asia, Middle East, Africa, we are starting to see some very, very interesting things, honestly, and there's a lot of overlap intersection between where agentic is going, where stablecoins are going, so that's keeping us super excited.

Rishabh Jain:

So you mentioned that you're seeing a lot on the bleeding edge of logged-in environments, and I want to use this to tie back a little bit to this idea of brand, potentially from a slightly different angle. One of the big questions that happens a lot with technological shifts is, is this shift more advantageous for independent merchants — so thinking a Shopify merchant — or a large aggregator? Ben Thompson does a lot of writing about this on Stratechery where it's like, hey, the walled gardens will see advantages in this wave from this shift, most recently the privacy shifts, he was worried about aggregation happening a lot. The way you stated your observations, it makes me wonder the same question, basically. Are we seeing advantages start to — technological advantages — accrue more to the walled gardens, as it were, or the large apps who have a captive audience? And then underneath that, it suggests to me that maybe this brand question that you were asking is actually ... I mean, it makes it even a harder question, right? Because basically, you're entrusting an agent in the context of a trusted application. The importance of the brand in that transaction is even harder to understand, basically. How are you seeing that emerge and how do you think about that?

Rubail Birwadker:

Honestly, Rishabh, the bottom line is, I have no idea, but I'll tell you the facts. The facts are that today, if you are a small merchant that has an enormously good product or a good service that you have to offer, your best way of getting eyeballs and consumers to use it is using advertising of some way, shape, or form. You're using a lot of word of mouth, you're doing influencer marketing, you're on X, etc., etc., etc. Now, in the agentic world, some of this will still exist, but one thing is for sure: We're going to see at least how advertising works evolve and what would that look like. So it's really about what tools are going to become available for those merchants. And again, like I said this earlier, which is I think again, there's going to be a cottage industry of companies that are going to be built to make sure that people with the right talent, stuff that they're building, services they're providing, get enough access to the buyers from around the world.

So I don't know what that looks like, honestly. What I do know is, whoever is smart enough to understand where advertising is headed is probably ... which, it's like, who knows? It's a complicated question. But on the other hand, which is where brand ... but on the other hand, it's sometimes hard to get found, even today. You might have the best product and it's very, very hard, even though you can buy ... How much money can you spend on AdWords and Facebook advertising and things like that? And ultimately, if you truly have the best product or service, you would expect that, in a free market, all these smart agents are going to get to the best answer, the best provider.

Your hope is that the intelligence of the technology that is being built will always get to nearly the right answer barring other exogenous variables that could prevent it. So in some ways maybe it's an exciting time because now, if truly, maybe quality shines over anything else, and maybe they have an easier way to get discovered because agents will figure out like, oh, this is the best product because I did all this algorithmic math in the middle and maybe it's something obscure in whatever, the southeast village in Japan, but it's the best thing that a consumer is really looking for, which ordinarily would have a very, very hard time surfacing itself to a buyer in San Francisco. I don't know. I actually really see it both ways. I really don't, I really don't know. And I think about this problem a lot.

Mike Duboe:

So Rubail, I know we're approaching time here and one topic I wanted to make sure we spent a minute on was looking at this more through the eyes of a consumer. And so we're spending a lot of time here talking about all the infrastructure supporting this wave, but actually for a consumer, let's remember at the end of the day, shopping is also a fun experience in many categories and we're now creating a new modality to do that. What kind of experiences do you expect consumers start to really benefit from and maybe what are some of the more interesting examples that you're seeing right now of agentic commerce through a consumer's eyes? What are you most excited about on that front? Or what should consumers be most excited about?

Rubail Birwadker:

We do a whole bunch of consumer research very, very regularly to just do a little bit of a temperature check from consumers, saying, What are you thinking? And the same consistent themes have showed up in almost all the research we've done, which is safety, security, I want to make sure my money is safe. That's a fundamental building block. The second piece, we are obviously seeing a little bit of early thinking where we think a lot of ... It's really where there is inefficiencies in buying. I actually think agentic commerce is going to make it easy. I think a very good example, one of my colleagues gives me all the time, which is he's got four kids and every time he has to go on a family vacation, which is twice a year with his four kids and his wife, he's like, That's a four-hour project for me.

I get on and I look at all the comparison of the hotels and all the comparison of the airlines and I look at all the various locations and the weather and all the activities my kids could do. And that has now gone from four hours to maybe 30 minutes for him because all he does is he prompts and then he makes decisions based on the recommendations, and he prompts and at least gets to the right answer. I think one thing is pretty clear: The early discovery process — at least the early funnel, first 80% of the funnel — that's just collapsing very, very rapidly, which ultimately is the precursor to anything that happens in commerce. We know that, for sure, is going to collapse, which is where you have to do the work to which restaurant to go to, what kind of dishes do they have, the kind of vibe a certain bar has or a weather a certain place has.

Just the individual work that went in to do that is going to just collapse by, I think, is going to collapse pretty rapidly. So that feels like some very, very early categories. And then there's a bunch of purchases that we do but we don't really want to do. It's just there because we have to do it, just paying bills, just go to my bank account and do it as long as it hits certain thresholds like paying rent, like paying tuition. And I just think there's going to be an enormous amount of chunk of purchases from consumers that consumers have to do but don't really want to spend the time doing. I think are very, very early, I think are going to be very early use cases where we'll see a lot of adoption of agentic commerce, especially once it starts to solidify from a trust perspective — a lot of it will come down to trust, the sooner consumers start to get trust.

But again, there's going to be always a classification of purchases where consumers don't want to use agentic in the same way that there's physical commerce, when eCommerce came along, physical commerce didn't go away. A lot of the world is still physical commerce. I don't always buy my tickets before. I sometimes walk into an AMC and I buy my tickets on hand. Sometimes I do like to peruse and walk around a Ray-Ban store or whatever else it is, and some of that's just not going to go away. I think if I was, I don't have a crystal ball, but if I was to make a prediction, I think you'd see a percentage of the world of commerce move to agentic, but the other channels are going to exist and they're just going to evolve and we'll just have physical commerce, eCommerce, mobile commerce, app commerce, and you're going to see agentic commerce. And it's going to be cool.

Rishabh Jain:

Amazing. Okay, so to wrap us up, Rubail, and I know you've mentioned a couple of times your crystal ball, I understand just like mine, is very hazy. Predicting the future is super hard. We just have to see what's in front of us and then allow for the future to emerge. From that perspective, what are some of the things that you're most excited about that are either new things that you're seeing either released by Visa — I know you guys just had this MCP release — or that you're seeing in the startup ecosystem or with your bank partners that you feel like are forging that pathway that allow for that future to unfold. And then to the extent that you're willing to, what pathways are you most excited about that are likely to emerge from some of these innovations that you're seeing?

Rubail Birwadker:

Yeah, look, I think on one hand I'm very excited as a consumer and a voracious user of all things new technologies, I'm very, very excited. I hope loyalty really becomes real. I don't know if you guys remember, 10–15 years ago, people were like, Oh, you'll be walking outside a Starbucks and you'll get this ping to say $5 off your next whatever, Frappuccino. And that just never really came to pass. Just closing the loop on attribution, data, personalization, never really came to pass. As a consumer, I still struggle a little bit with loyalty Yeah, there's airline loyalty, but almost everything else, it's kind of janky. And it's because the closing of the loop is very hard with context and data and what I really want and how does it show while still keeping privacy front and center. So I just don't [think] the world has really solved that problem.

So actually as a consumer, I'm very, very excited. What does that mean in the agentic world? I really, really hope people go build awesome things in agentic that make me a more loyal customer, but also gives me a much more enriched experience — buying experience — because that's just enjoyable to do. Putting my Visa hat on, look, I think the most important thing we're excited about is, we want to roll out our agentic commerce standards to as many markets in the world in the coming year and we want it to work. I think I will feel very accomplished if we see tens of millions, if not hundreds of millions of users using the Visa cards to buy things in an agentic environment at better or lower fraud than they do today in an eCommerce environment. We'll feel very accomplished because ultimately our job is to digitize the world and let consumers pay and get paid anywhere with our credential. And we really, really believe that we are working very, very hard at that. And look, everything else is bells and whistles. The bread and butter is how do we make consumers to purchase in a safe, secure, fraud-free environment as broadly around the world as possible? And agentic is a new challenge we're facing on that.

Rishabh Jain:

Amazing, Rubail. Thank you for your time. This was amazing.

Rubail Birwadker:

Thanks guys. Really appreciate you guys having [me].

Mike Duboe:

Thanks, Rubail.

‍