Building Trust in the Age of Agentic Commerce (Episode 166) #DataTalk

Speaker A: Hello, friends. Welcome to Experian's Data Talk podcast and video show, where we take a deep dive into key data and analytics topics with industry leaders from around the world. This episode is part of our special limited series, Fraud Frontlines. Focusing on the latest fraud trends and insights from seasoned experts throughout the series. We're excited to have Kathleen Peters, uh, our Chief Innovation Officer at Experian, as our guest host. Let's get into the episode.

Speaker B: Welcome to DataTalk. I'm Kathleen Peters, Chief Innovation Officer of Experian Identity and Fraud, and your host for the Fraud Frontline podcast series. Today I'm joined by Farooq Sheikh, Vice President, Global Industries and Solutions at Cloudflare. Cloudflare plays a critical role in protecting the Internet and increasingly, AI agents that operate within it. Cloudflare is an important collaborator in Experian's recently announced Agent Trust framework. Farouk, welcome to the show.

Speaker C: Kathleen, great to be here. Thanks for the invitation.

Speaker B: Wonderful. So to get us kicked off and started, Farooq, I'm hoping you can introduce our audience to yourself. Tell us a little bit about yourself and your role at Cloudflare.

Speaker C: Very, very excited to be here, Kathleen. As you mentioned, my name is Farooq Sheikh. I lead global industries at Solutions at Cloudflare. Cloudflare is a Internet and AI, uh, infrastructure company. We spend all of our time trying to improve, protect, connect Internet and agents now, including providing security as well as runtime services to AI applications. My focus is working with our most strategic customers across each of the industries, whether it's financial services, whether it's media, whether it's retail, on how we can bring together some of the newest and coolest capabilities. We are very excited about Agent E Commerce and all the work we're doing together. Kathleen, I'm super excited to be here.

Speaker B: So. So, as you mentioned, we recently announced together Agent Trust, that first of its kind framework to bring verified human identity to these agents which are acting increasingly autonomously on our behalf. So from your perspective, from Cloudflare's perspective, why is this announcement and this collaboration that we've done so timely right now,

Speaker C: from our vantage point, as I, uh, mentioned, we support more than 20% of the Internet. So we see a lot of the activity of the Internet and it's clear to us that we are in the middle of a very significant platform shift. If you go back in time, we went from desktop to mobile to social, and now we are transitioning to the AI era and the Internet as a whole is changing. What people do on the Internet is changing how they do. It is changing as well. So in that context, what we are seeing increasingly is that trust layer that was needed for commerce to emerge, whether it's E commerce or local commerce, where you have a trusted kind of counterparty, that trust layer needs to be built for agents. As AI and agents start spending more and more time and have a bigger role in commerce or agentic commerce, you need to rethink and rebuild that layer for the Internet, for, for AI. That includes thinking about how websites understand who's a trusted agent, what agents can do on a website, how they navigate that website, what information they can provide, how do they address questions around identity, around trust, around many other dimensions. And all of that is critical for the next kind of platform shift, for everyone to adopt AI. So we are very excited.

Speaker B: I'm so glad that you mentioned that trust layer. I know that as I've had conversations, we've talked about this before, I know you've had a lot of conversations with the various parties in this space and even consumers. Even as I talk to my family or my co workers, my friends, and I'll ask them, have you tried buying anything online yet using an agent? And people said, oh, you mean searching? I'm like, no, actually buying something. Have you given that a try? Most of the time the reaction that I get is, no, no, no, I don't trust that. And, and I feel like we need to find ways to help consumers build that trust. For merchants and retailers to build that trust, there's the card networks that are working to help assist in that trust, to trust the transactions that are presented. Cloudflare obviously plays a really important place in that trust layer as well.

Speaker C: It's exactly that. It's like four different stakeholders, right? Like, if you think about, and the concerns are so different, right? The brands and the retailers really care about how the agents engage with their site. If you, if you go back in time, a lot of these brands were trying to protect, um, their websites from, you know, bots that were trying to scrape their, uh, inventory or creating security risks. So now they have to look at these agents as a source of commerce.

Speaker B: Right.

Speaker C: And being able to distinguish bots, which may be malicious, versus agents who are trustworthy, carry some form of, um, identity and payment information that are representing a customer. That's like one of the big things that the brands have to, to do, right? From a customer perspective, a little bit of the concern historically has come from, hey, whether the agents will do the right things, will they follow instructions, will they hallucinate, will they go to A brand that I don't care as much will they execute on my kind of requests and requirements. Precisely. And so forth. And that's a whole kind of set of things that agents need to get better at doing and E Commerce needs to facilitate. And from the card networks, the big issue there is around, like, how do you think about refunds and where does liability sit? Right. If an agent buys something that may not be exactly, uh, what the customer wanted, but the transaction happened, how do you facilitate that trusted exchange? Right. And if there's a fraud element, it's all new space. You don't know how fraud will occur in agentix spaces yet. Right. Like we don't know yet. We're learning collectively how do you think about that and how do you manage that? And so it's a set of space where we have currently more questions. But the good news is that there's a large number of people who are very actively working on addressing these questions and we're very happy to partner with Experian on that journey.

Speaker B: Yeah. So there is a lot of energy around identity, identity verification, identity. Um, for agents, certainly that is very critical. That's why we knew that we had an important role to play here and bring our experience to bear as well. I like how you were talking about transactions and when transactions may go wrong. Because even legitimate transactions today, obviously there can be mistakes. Uh, there can be the wrong thing shipped, there can be the wrong thing ordered. We need a different size. But then there's also the potential for fraud. One of the things that I believe has built trust in the way that E Commerce transacts today is that all parties have a general and accepted understanding of what happens when something goes wrong. We understand what the process is, who's on the hook, who's responsible, who's accountable, what evidence will need to be supplied from a merchant that this is a chargeback, which is legitimate or not. And then that kicks into gear, the various processes as to who will ultimately be liable and responsible. So many of those historic checks and balances have been related to knowing who that human is and what their intent is. To your point, we believe that having that as a tokenized part of the agent itself bound to a human with that human's intent, with their confirmed payment information, is going to help build that trust so that more of this potential and these sales can scale over time. I think 100%.

Speaker C: Right. Like I, when we look at this, the, the problem statements, they're few. Right. So we think about, like, as I said, right. Like, what do Consumers want, what do merchants want, what does the payment ecosystem want, and so forth. Right. And the broad kind of dimensions where we're spending a lot of time is around number one is interoperability, which is, hey, you have existing payment reels, you have some new payment reels. How do we make sure that consumers have access to all of them? Then? The question of trust and identity is a critical one. There's a lot of work to your point, like has that has been done in the human identity space, like validating identity from kind of the all kind of capabilities you guys provide and how do you extend that to the agent space and how do we connect the agent identity to the human identity and make it easy for the issuing banks as well as for the merchants to correlate that this agent is actually fireworks agent and firework is a customer and has a loyalty kind of association with that merchant or brand or retailer. Right. And that kind of leads into the third issue, which is all about disintermediation risk from a brand perspective. They care about loyalty, they care about capturing the buying behavior and the activity and relating the agent to the customer that they know well across what used to be omnichannel and now it's agentic and omnichannel, uh, conversations.

Speaker B: I hear that from merchants a lot. The same where they want to make sure they can maintain that relationship both to ensure a great customer experience, but to build that relational ship over time so that they can make additional offers to that shopper, maybe make suggestions or even assist them with the shopping. They want to maintain that relationship and not have their inventory and their experience commodified, basically. You also talked to Farouk about, and uh, mentioned obviously Cloudflare does power and protect a significant portion of global Internet traffic, upwards of 20% or more. Why is that position so important? When we think about what it takes to secure AI agents broadly, and I

Speaker C: think that position is very important. But even more important is our mission. So Cloudflare's mission is to help build a better Internet. That's kind of what we are, uh, focused on. Uh, one outcome of that mission is that we have built and provided a lot of infrastructure and tooling and security capabilities to nonprofits and folks who are running elections and all these places where Internet was the least secure. Right. That's the heritage with which we come to this conversation. Now, if you look at fast forwarding this conversation to a, uh, place where there's more and more agents operating transactions on the Internet, security and trust is critical. Right. So for example, if I'm trying to buy um, a particular product. I might look at two or three sites, might look at review, um, from a few different vendors and buy a particular product. But when an agent is trying to do that, that's a very different approach. They might look at 50,000 sites. Right. Or 50,000 reviews. They might try to do things which are a lot more intensive and that creates an overhead at the Internet level which we care so deeply about. Right. Both in terms of security, but also in terms of compute and performance and so forth. Right. And that's where we are spending a lot of time getting to a place where we can provide the same security, trust, um, and cryptographic layers to an agent. So when an agent comes to a, ah, brand or a retailer, it's verified through webbot Auth, which is a cryptographical signature that tells you that this agent is exactly who they say they are. It's not a malicious kind uh, of representation. It's carrying a verified payment credential, verified identity based on all the work that uh, Experian and your team has done and that it should be allowed to pass. Right now that's where we are trying to spend most of our time, uh, with our partners.

Speaker B: I know you're spending a lot of time thinking about the right architecture too, to make sure that this is performant, that it provides a really great experience for shoppers. How is Cloudflare thinking about really enforcing that trust at the network edge again?

Speaker C: Right. Like our job here, given our mission to help build a better Internet, is to provide the tools like what we, we are in the business of providing transparency, control and trusted kind of infrastructure to our customers, and our customers then decide if they are open to agentic transactions. Right. That's kind of the, the basic premise of how we operate. In that vein, we think that if we continue to build with our, the trusted infrastructure, it'll be easier for every merchant, every retailer to adopt this capability and for the industry to adopt this capability. And that's good for the Internet broadly. That's where we're spending a lot of the time in terms of how it manifests itself in execution. Cloudflare has more than our points of presence in more than 300 cities. Uh, we, as I said, serve more than 20% of the Internet population. And what that does is that every request that runs through Cloudflare can be, you can enforce that layer of security at the point, at the local point of presence globally and it can validate, uh, based on the work that we've done together, that the Agent has a uh, uh, authenticated identity provided by and secured by Experian. That we think is a good kind of starting point for this conversation.

Speaker B: I love that angle that you mentioned on trust too. And that's a new one that I hadn't thought about in that if developers know that they can manage the risk and they have trust in that security, that they'll feel more freedom to innovate and create and think about new ways and new use cases to use that technology. If they feel, if they trust the security of the tools that they're using and the networks that they'll be transacting

Speaker C: over and the card networks will facilitate the transactions and the PSPs will get more involved like it, you need financial services as a trust, um, based conversation. So that's where most of the work is.

Speaker B: That's true. It's true. So we've been talking a lot about uh, agentic commerce and building trust for agentic commerce. Obviously being able to understand the human, the intent, the source of agents has broad applicability both in the enterprise and we think about workforce agents as well as other consumer use cases. How do you see that landscape in terms of the applicability of uh, some of the things that we're working on together even?

Speaker C: I think the thing that excites me is when you like to your point, right in the beginning you mentioned that hey everybody or more people have agents now, but when you ask them, hey, are you, is your agent able to buy something for you that's still kind of uh, a maybe to a no kind of an answer. And what I'm excited by a little bit is like hey, we are not that far from the time when there's an inflection point and more and individuals are able to uh, buy more things via agents. Now there's a whole debate here around trust and identity which we were talking about, but also around what are the kind of things that people might buy, whether it's travel, whether it's uh, um, kind of day to day consumer goods and so forth. Right, that's, that's great, that's good. But the thing that excites us is that I think we are starting to build the rails and the guardrails for agentic commerce where you both have the ability for agents to carry trust, identity and payment information, but also be able to have guardrails around it from a security perspective and get merchants comfortable and card companies comfortable. So I'm excited that this infrastructure is going to um, convert into more and more people getting comfortable with Agentic commerce. I think we're a little early, but we're not that far.

Speaker B: I agree, I agree. Well, I mean we do a little bit of crystal ball looking ahead too. I think it'll be interesting to watch the end of this calendar year as we enter the annual holiday shopping season to see how consumers, retailers react, what new things we might see. I think we have a bit of a window here between now and next shopping season. How do you see moves and plays maybe between now and the end of this calendar year, but then next year as well.

Speaker C: I think last year was the first time we saw referral traffic from Agentix services become a very significant portion of the overall referral service. Right. So it's still low in terms of the overall traffic to retailers, but it's a much higher percent of overall referral, uh, traffic, which I think was a very surprising scene. Right. Because like agents at that point in Agentix services were less capable, but you were still seeing that people were engaging with those services very regularly for like, hey, what gift should I buy? Or what should I buy in this scenario? That whole discovery journey. Exactly. Discovery was very, it was very clear that discovery is moving at a very significant pace to various AI surfaces. And if, and again, crystal balls are tough. But if I think about rolling that forward, as long as we have sufficient trust in that. So we didn't have enough trust infrastructure in place last year, but I'm excited about this. Yeah. Like this year I'm hoping there's m more to be done, but if we get to the right structure there by which we have the trust infrastructure discoveries already in AI surfaces, I can see us getting to a place where more of the transactions are going. And then to your comment about the year after, I would expect even more. I think there's only one direction here in terms of more and more agentic transactions. It's just we need to build the infrastructure to make it happen.

Speaker B: I think you're right. It's obvious that shoppers want the convenience. They're curious about this. As people are getting more familiar and more comfortable with asking for suggestions, asking AI to find them a good deal. Right. I think that that discovery element is going to be expected. And then why wouldn't you want to take that next natural step of yes, please, just go ahead and get that for me. I'd like it shipped right away. So I think consumers will be pushing the envelope as much as the overall ecosystem will support it. I agree with you. And it can only go up from there. I think also too that Fraudsters will teach us where the risks are, where the vulnerabilities are. We'll start to see those, right? No, no, I think that that's going to, I think we're going to see that happening.

Speaker C: I worry when we don't see enough edge case scenarios come up. So I think we will see a lot of those edge cases, uh, as we go into the reality of this.

Speaker B: Yes, as things mature and as more people are experimenting and I think as ecosystems are coming together like ours as well, we will naturally begin to look at more of those edge cases. It's my belief that there have been a lot of what I'm calling happy path technological demonstrations as protocols have been announced and we can show that, yes, this bit can connect to this bit and result happens not so much on what those edge cases are and when things aren't so happy path, as we discover those, we can put the right things in place. But I think it is really powerful and significant that large players like ourselves are coming together to say, how do we make this safer? How do we make this more trustworthy? How do we bring together the systems that will provide that accountability that will lead to trust and to greater adoption? It's a really great time to be a part of this.

Speaker C: There's not a lot of uh, platform shifts that happen and it's also, it's more rare to see a platform shift of this dimension happen. Like case in point, I think, yes, going from desktop to mobile changed payments, but like not as much. Right. Like E commerce changed and mobile apps became more important. But payment was payment. Right. I think this one is really significant in terms of how significant a, uh, platform shift, this AI revolution is and it's really impacting everything. We're very excited.

Speaker B: Farooq, what advice would you give to organizations who are just beginning to prepare to support agentic driven transactions? If I'm a financial institution or a merchant or retailer, what advice would you give as they're thinking about how to get started?

Speaker C: I think the way we think about it is that um, AI is a contact sport, right? Broadly, like you, you need to be involved, whether it's experimentation, whether it's trying things out, learning like whether it's the protocol side, the transaction side and so forth. It just, it's not a theoretical or like concept that you can easily absorb without getting your hands dirty. So my biggest advice to every organization is like, have a thoughtful approach to how you experiment and engage with AI, uh, AI driven transactions. How you think about adopting your systems and people and processes to enable these capabilities because this is not a, this is a significant platform shift, but it's also very involved effort.

Speaker B: Any use or scale deployment of AI definitely is more than just the technology and the platform. To your point, it's change management, it's ways of working, it's skill set. I think as businesses are thinking about how they transform themselves, it's more than just the technology, it's the people, it's the ways of working, it's understanding that change and managing it in terms of overall organizational effectiveness.

Speaker C: That's right.

Speaker B: Any other news or highlights from your world or things that maybe are keeping you awake at night that you'd like to share as well?

Speaker C: No, I'm just excited that uh, there's one less thing, uh, for me to do because I can have my agents do my uh, you know, everything from my weekly groceries to all the way to my Christmas gift planning. So I'm just excited. This is great.

Speaker B: That's fantastic. Well, Farouk, I want to thank you so much for joining me today and to talk about.

Speaker C: Thank you.

Speaker B: AI agents, agentic commerce and building that trust between consumers and retailers and the ecosystem in between. As AI agents are becoming more prevalent participants in E commerce, it's clear that these kinds of collaborative partnerships are going to be critical to provide that trust and safety, um, for everyone involved. And I'm excited that we get to work together to make that happen.

Speaker C: And thank you Kathleen and the Experian team broadly. We really enjoyed the partnership. So really appreciate all the work uh, that we have done and all the work that's underway. Uh, the big uh, like if you think back to our mission, our mission is to help build a better Internet. And the word help is there to signify that it's not a job that any one company, um, people can do on their own. So we really appreciate the nature of the partnership and how we're helping to build a better Internet.

Speaker B: Wonderful. Thank you for our listeners. If you'd like to learn more about Experian Agent Trust and Cloudflare's role in securing AI driven commerce, check out the links in our show notes. And as always, thank you for joining us on Experian's Fraud Frontline podcast series and we'll see you next time.

Speaker A: Thank you. Thank you for listening to this episode of Data Talk. As a quick reminder, our show notes and video clips from today's podcast are available on the Experian blog. Just go to experian.com datatalk where you can also find the full archive of past shows. If you enjoyed today's show. Please consider sharing with a friend and giving the show a five star rating. Your vote can give this show more visibility.

Speaker C: Sam.