“Taking the hardest path in payments”: How Adyen’s new AI platform boosts conversion while reducing fraud

Artificial intelligence is rapidly transforming the financial landscape, creating new opportunities for innovation while reshaping consumer expectations around speed, personalization, and security. As financial institutions and businesses race to integrate AI capabilities, we’re witnessing a fundamental shift in how payments are processed, how fraud is detected, and how financial services are delivered.

Today, I’m delighted to welcome Davi Strazza, President of North America at Adyen, a global financial technology platform that’s at the forefront of this AI evolution. Davi leads Adyen’s North American operations, where the company is leveraging artificial intelligence as a strategic growth driver through initiatives like their recently launched Uplift platform.

In our conversation today, we’ll explore how Adyen is implementing AI solutions to enable hyper-personalization, enhance fraud detection, and automate complex financial processes. Davi will also share his insights on the state of real-time payments in the US and what companies need to do to make instant payments a more fundamental part of our commerce and financial system.

Listen to the episode

Subscribe: Apple Podcasts I SoundCloud I Spotify

Watch the episode

AI as a strategic problem-solver, not just a trend

Adyen’s approach to AI isn’t about implementing technology for its own sake—it’s about identifying specific industry challenges and using AI as a solution. Strazza emphasizes that businesses should resist the temptation to adopt AI without clear purpose.

“I think the approach we’ve taken is to, instead of kind of make AI the end goal… we ask ourselves, okay, what is really our end goal as a business, right? And if we work backwards from what that is, what are the different areas and spaces where AI can make us so much more productive and effective. And so it really becomes like means to an end, like you need to start with a problem, and that’s what we did.”

Data scale as a competitive advantage

Adyen’s platform processes trillions in transaction value and has seen over one billion unique shoppers, creating a massive dataset that powers their AI initiatives. This scale provides unique insights that competitors can’t match.

“We have seen over 1 billion unique shoppers in our platform. So essentially, again, one in every five or six people that are buying products and services on Earth. We’ve seen that in our platform. So you combine, again, the one point the trillions of dollars, the 1.3 trillion euros that we processed last year, with 1 billion individuals in our platform, it’s a data set that allows us to very uniquely separate good shoppers from bad shoppers.”

Payments evolution toward invisibility and recognition

Strazza envisions payments becoming increasingly invisible and friction-free, with AI enabling better customer recognition across channels and significantly improved transaction success rates.

“I think further down the line, payments is really going to be invisible… I think payments is going to go more and more invisible over the years, and consumers will be much more recognized. And there will be great loyalty programs. There will be great applications for fraud prevention. I think the industry has a really interesting opportunity to see like step change in terms of performance and customer recognition.”

The big ideas

1. The holistic approach to payment optimization: Adyen’s uplift platform tackles conversion rates, fraud, and payment costs as interconnected challenges rather than isolated problems, allowing merchants to find the optimal balance between seemingly competing priorities.

“Sometimes we advise our customers to take more fraud for the benefit of much higher conversion and ultimately better bottom line results. And that is really like the holistic approach that we’ve taken with Adyen Uplift.”

2. Direct connections yield superior data: By building direct connections to networks and issuers without intermediaries, Adyen accesses higher quality transaction data that enhances their AI capabilities.

“We built direct connections to the networks, direct connections to issuers, in many cases, to the source of funds, the data that we get back from transactions is of the best quality. We gain unrestricted access to the data in the ecosystem because of our direct connections.”

3. ROI-focused payment technology: Adyen’s business model hinges on delivering multiples of value compared to their fees, positioning payments as a profit center rather than a cost center.

“For every dollar that a merchant pays us in fees, we we made it our mission to return multiples of that dollar in revenue to our customers, in bottom line results to our customers.”

4. Cross-channel customer recognition: Current retail environments struggle to recognize the same customer across online and in-store experiences, creating a significant opportunity for AI to bridge this identification gap.

“When you go to an e-commerce site, you leave something behind, like an email, a phone number. When you go to a store, you check out, you leave nothing behind. You are a ghost.”

5. Payment ecosystem diversity: The future of payments will involve multiple competing technologies and rails, with payment processors adding value by abstracting this complexity for merchants.

“I think there will be competing payment methods, there will be competing rails, there will be competing infrastructure. And that’s really positive. We really welcome that shift… I think our role is to abstract the complexity of dealing with all of those different pieces of technology.”

Read the transcript (TS Pro subscribers)

Understanding the Uplift platform

the right way to think about uplift is it kind of tackles some of the largest problems in the payments industry, and those are conversion rates. So essentially, how many transactions are being approved? It tackles fraud, which is pretty straightforward, and cost, cost of payments specifically so, how much does it cost for a immersion or a business to take a transaction? Those have been challenges in payments ever since. Again, we we saw the first car transaction coming in, like decades ago, and I think over time, the industry has taken many different measures to try and tackle those issues. But the vast majority, if all of, if not all of those measures have been taken in isolation.

The difference with uplift is that it really helps merchants, specifically, enterprise merchants, find the right balance between all of those variables and get to outcomes that truly benefit their their bottom line results. And so let me, let me unpack that. Yeah. Still, if you Yeah, if you think about a transaction flow, there are a number of different steps right between the moment a customer is browsing the website and searching for their product of choice, they go to the checkout, then they click Pay. That transaction is submitted for processing. It goes through some sort of risk screening all the way to until it hits the issuer of that card or the owner of that payment instrument for approval, right?

And so, because of the fragmentation in the industry, Zach there are many different companies and many different pieces of software that are touching that transaction, and there’s been several different attempts to kind of optimize every single step of that process of that journey, and what we saw was that because those steps were being taken in isolation, sometimes businesses were left with results that were not necessarily great for their bottom line.

An example, if you, if you have a big fraud issue, and you you want to take your chargebacks down to zero or to the minimum possible. You may, you may create rules that become so strict that will hurt your conversion rates, but if your systems are not talking to each other, you’re not going to know that. You’ll be happy with your chargeback results, and you’re going to be blind in terms of the impact. The impact it has on conversion rates and vice versa. Sometimes merchants are so focused on growth that they lose track of how chargebacks are evolving, and that can cost them millions of dollars in any given year.

And so it’s interesting that when you start to look at things in context and in perspective, you get into interesting discussions where we advise our customers to take more fraud for the benefit of much higher conversion and ultimately better bottom line results. And that is really like the holistic approach that we’ve taken with Agent uplift. There’s a lot of you know. Details that go into the sausage making. We can go there if you if you want me to, but this is what really makes uplift stands out. It looks at the entire purchase funnel, and it uses our entire like aI platform to kind of optimize for the bottom line results.

The power of network data

Absolutely. So there are few things that make uplift very powerful. The first is it’s it’s built on trillions of dollars in transaction value that we processed over the years. And not only that, but because at agin, we built direct connections to the networks, direct connections to issuers, in many cases, to the source of funds, the data that we get back from transactions is of the best quality. We gain unrestricted access to the data in the ecosystem because of our direct connections. And so as far as like data integrity and quality, we are probably the player that is best positioning the industry to kind of capture that right. And so that is number one.

The second is because we operate across online in store, because we work with marketplaces, we work with SaaS, platforms that are embedding payments. We also help businesses with onboarding. We also help businesses with payouts. And so we get access to other data points that are very interesting and important when it comes to identifying a individual. For example, if you are onboarding a if you’re a seller in the marketplace, there’s some KYC that is needed. And you may share a passport, you may share an ID to onboard. And so that is a very important piece of data that we also collect in our platform.

And so when we connect all the dots, we are in a position where we can truly recognize customers as they come in, as those transactions come in, right? And over the years, I mean, we’ve been doing a lot of research, and we’ve been investing a lot in identifying those customers. The bottom line is, we have seen over 1 billion unique shoppers in our platform. So essentially, again, one in every five or six people that are buying products and services on Earth. We’ve seen that in our platform.

So you combine, again, the one point the trillions of dollars, the 1.3 trillion euros that we processed last year, with 1 billion individuals in our platform, it’s a data set that allows us to very uniquely separate good shoppers from bad shoppers. And so if you are a business using agent, we’re going to look at that transaction. We’re going to look at all the different pieces of context around your business, around that shopper. We’re going to cross reference that with our platform wide data, and we’re going to be able to tell again, is this a good shopper? Is this a bad shopper?

If there is a question in terms of which one is it, we can also kind of implement tools to help you fast track the transaction or not, put the right, the right risk processes in place in order for you to get again, to bottom line results that make sense for you according to a strategy and so and so. Absolutely, I mean, data is a big part of uplift. Uplift is not possible without without the data that we have gathering the platform.

Technology evolution behind Uplift

Yeah, that’s an interesting that’s an interesting question. I think, in terms of accessing data. We took the approach of building agin, connecting directly to the networks, and so we never really had a problem in terms of accessing data right. And in fact, again, we are probably one of the only businesses of our scale that has has built a global platform the way we did, without any third parties, without any MNA and intermediaries, again, that gives us an edge in terms of capturing data.

I think what changed in the past years truly is really technology in terms of how you process that data and digest that data and then translate that data into insights. And look, I think AI really is the fundamental technological breakthrough here that that kind of helped us take that deep in terms of how we digest data and translate that to insights.

We have always been really big on connecting the two. Words, online and offline, and looking for, again, all of those connection points to kind of make sense of the data in our platform. Ever since, again, we started to invest in AI infrastructure and train our models in house and such, which, again, it’s been a few years down the line. We we found that in doing so, we build the ability to learn from every transaction and dynamically kind of route that transaction in the best possible way. That is something that previously was done more in the ad in ad hoc, ad hoc basis. Now we can do that dynamically. That was really the big leap that happened in terms of how we optimize payments and dynamic routing, thanks to the dynamic routing that is really thanks to the investments that we made in AI, and how we train our models in house and build teams around it to kind of help us productize uplift, which is what we did and launched the product in January.

The strategic investment in Uplift

Oh, absolutely, it’s, it’s incredibly strategic. I mean, most of the teams involved in building uplift are engineering. I mean, half agent is engineering, and many of those teams, engineering teams are essentially data scientists that are helping us again, build those building for structure, trainer, models and productize uplift to the point that is what it is today.

We think that. We think that, generally speaking, like payments is only getting more complex, between different payment methods, between different features and functionality, between regulatory changes, the space is getting more and more dynamic, which, in turn, kind of creates complexity and challenge for businesses, and really for us, kind of opens up opportunity to help businesses right, help our customers.

And so when we when we look at what uplift could do and how we could help merchants with these big problems, conversion, cost and fraud and and we saw again, the early results that we were getting like off the bat, we were very confident that this could be probably one of the biggest breakthroughs in the company and the biggest product launches that we had in the company. And we’re seeing that right now. We run a program with 60 merchants, 60 of our largest customers globally. They ran uplift for several months in a row, and the results we saw from that program are just amazing. 6% of authorization rates, uplift in average. That is massive. That is millions of dollars a year for enterprise business while seeing the cost reduction of 5% in average as well, which equates to, again, another few million dollars a year.

And so if you, if you look at those two elements right, and you combine that with the easy of implementing uplift, is literally a toggle in our customer area, it becomes a very powerful product and and in our customers have been giving us great feedback. There’s a long way to go for us to kind of push the product to our entire platform, but the adoption has been great, and we’re very confident and happy with the results.

Uplift’s business model

Yeah, it is an add on for our merchants technology. From a technical perspective, it’s really simple to turn on. A lot of the work, a lot of the routing and the optimization, takes place behind the scenes. I think what we what we see with with uplift, is very similar to what we see with pretty much every other product at agile. I think we, we are on the mission to help customers kind of really take strategic value from payments, and so if you look at how that manifests for every dollar that a merchant pays us in fees, we we made it our mission to return multiples of that dollar in revenue to our customers, in bottom line results to our customers.

And so the ROI there has been just really incredible. Like I said, if you take an enterprise customer and if you increase their authorization rate rates by 6% if you help them reduce costs by 5% that is like 10s of millions of dollars in a year, and that outweighs the urgent fees by multiple. Goals and so.

Innovation and differentiation

Oh, absolutely. But again, I think we, we, we’ve made the decision early on that we would take the hardest path in payments and by building the platform from the ground up and and connect directly to networks and and issuers and and, and that puts it in a position where we can innovate, we can push these products. Uplift is one of them. Tap to pay is another one. We’re leading the space in US debit routing as well. We’re very proud of those results, but we are specifically proud of the impact that those products have to our customers. That is what, really what drives us.

AI approach and implementation

Yeah, yeah. Look, clearly, AI is a transformational technology, right? And some people compare it to the internet. Some people say it’s bigger than Internet. I mean, whatever it is like, it’s really it’s really meaningful. I think with my way to look at this is with any new technology you kind of need to think about how, how you, how will you implement it, and what kind of results you’re looking for.

It’s very, it’s very tempting, right? I think many companies they, I mean, if they, if they look around, all they’re going to see is Ai, ai doing this, AI, doing that, AI, making everyone more productive. And so they get very tempted to start just doing whatever it is, to kind of implement AI into their business, to help their workforce, or something else.

I think the approach we’ve taken is to, instead of kind of make AI the end goal. Make using AI to end goal. We, we kind of ask ourselves, okay, what is really our end goal as a business, right? And and if we work backwards from what that is, what are the different areas and spaces where AI can make us so much more productive and effective. And so it really becomes like means to an end, like you need to start with a problem, and that’s what we did. Like, what are the big problems we’re solving for our customers, and how can AI help us? Right?

And frankly, we found many different areas where AI could help us achieve our goals. But if, if that was not the case, I think we’ve, over the years with with we’ve proven to be the type of company that would not just invest in AI, because everyone else is doing it. And so that’s how I look at it. I mean, when it comes to product, I mean, great product, great, great products. They are, they’re, they’re, they’re great because they solve big problems, because they generate great results. And so you have to start with that and work backwards from what makes sense for your customers. That’s the approach we’ve we’ve taken, and it’s been, it’s been working really well for us, right?

The future of payments

Yeah, no, absolutely. Let’s, let’s take a step back, like if you, if you look at e Commerce Industry, average, 15% of transactions on E commerce. They. Bail, they are declined, and they’re declined for many reasons, including fraud prevention, including, like, technical errors, including, of course, when the customer doesn’t have funds in their accounts, and so there’s just a lot of money that is left on the table. Like, payments is not a solved problem. It’s far from being a solved problem.

When you think about the retail space, one of the biggest challenges that retailers have is they still they still struggle to kind of recognize customers as they transition across their e commerce and in store environments. When you go to an E commerce site, you leave something behind, like an email, a like a phone number. When you go to a store, you check out, you leave nothing behind. You are a ghost, right? Just fingerprints, exactly. And so it’s very hard for retailers to kind of recognize that you are, you right and let alone build a profile, let alone, like build a loyalty program that is very specific to you, and those are just two examples.

But I think as we progress with the use of AI in the payments space and really broadly in the financial sector, I think we’re going to see great results in terms of performance. So again, think of a word where you no longer have transactions to the E commerce space declined because there was just not enough data to be able to tell if that was a good shopper or a bad shopper. We closed that gap.

But there’s also a word where you as a consumer, you are instantly recognized as you land on a checkout page or as you walk into a store. And so what that looks like in practice is you’re always going to find your payment method of choice at checkout. You’re always going to be welcomed with like suggestions and products that make sense to you, the kind of are in line with your history of shopping in that store.

And so I think we’re going to close many of those gaps that exist in the industry today, and exist because we haven’t really had the technology to help us kind of connect all the dots, right? And I’m talking about the data points in Commerce today, and so I’m very excited about it. I think further down the line, payments is really going to be invisible, like we we take Uber for granted today, but the reality is, there are many places where you’re shopping where the shopping experience is very clunky. It’s there’s just a lot of friction.

I think payments is going to go more and more invisible over the years, and and consumers will be much more recognized. And there will be great loyalty programs. There will be great applications for for fraud prevention. I think the industry has a really interesting opportunity to see like step change in terms of performance and customer recognition. Which is which is really exciting.

The role of competing technologies

Well, there are all sorts of things that are involved in that vision. I think one, one other truth about the payments space and financial services is it’s very rare to see a piece of technology that becomes so dominant that it will take all of the share in that market, especially if you look at global scale, right?

And so I think there will be competing payment methods, there will be competing rails, there will be competing infrastructure. And that’s really positive. We really welcome that shift as Aji, and I think our role is to abstract the complexity of dealing with all of those different pieces of technology, different rails, and make sure that, from a merchant standpoint, they they’re being really effective, that they’re meeting their customers where their customers need them to, and they’re doing that in a way that, again, allows us to focus on their business and leave all of the hard tech and hard ground work to us to deal with.