Q2 Holdings, Inc. (QTWO) Earnings Call Transcript

All right. Hello, and thank you all for joining us. I'm Josh Yankovich, Vice President of Investor Relations and Corporate FP&A here at Q2. [indiscernible] before we begin, some of the comments today may include forward-looking statements that are subject to risks, uncertainties and assumptions that could change. Should any of these risks materialize or should our assumptions prove to be incorrect, actual company results or outcomes could differ materially from these forward-looking statements. A description of the risks, uncertainties, assumptions and other factors that could affect our financial results or outcomes is included in our SEC filings. Except as required by law, we do not undertake any responsibility to update these forward-looking statements. With that, I'm glad to be joined again by Adam Blue, our Chief Technology Officer; and Hima Mukkamala, our Chief Operating Officer. This is the second conversation in our AI series. The first early in the year was a deep dive on our strategy and architecture. And today, we're focused on some of the questions we've been hearing most from investors right now. What is the rise of personal AI agents meaning for the role of digital banking? How is AI reshaping the fraud landscape? And where does Q2 continue to have the right to win. So with that, I'm going to go ahead and start with a question with Hima. So Hima, let's start with personal agents. Recently, with news rolling out. There's been a lot of momentum in the conversation around using agents to connect with disparate applications, including digital banking and what that might mean for those respective industries. So Hima, where do these agents actually integrate with respect to financial institutions, digital banking platforms? And where does it leave us as their digital banking provider?

Yes. Thanks, Josh. When we think about the evaluation of how users, consumers, commercial users interact with their financial institutions, personal agents are the next banking channel, we have had branch people used to go to a branch, we went online, web and then mobile. And so I look at this as a new front door and not just a replacement for a bank or a replacement for digital banking and a fraud platform. Our technology just like we have enabled these interactions coming through mobile, web and so on, it enables these agents -- personal agents to come and interact with consumer or commercial institutions, financial information. And that's the evolution we are going through -- we're going to go through the coming years. There are a couple of ways we think about this. A bank can keep the AI experience inside the channel it owns. This is what we do today where you can go log into the website, there are agentic ways in which consumers can interact with the bank or the back office. The second approach, which is emerging with news like application or [ TOT ], like applications is how do they support the channels they don't own? It could be a social agent, it could be apps like [ instinct ], which use messaging applications and iMessage app or a smart device or the Claude and the ChatGPT. And that can happen in a couple of ways. Through the platform, we expose what's called MCP, which is a new way of API integrations. And so we expose MCP endpoints on most of the digital banking stack. And these agents -- personal agents can use that to talk to the digital banking platform. The second approach in this subset is not just use MCPs because MCPs are just APIs and the banks want a richer experience and want us to facilitate making this agentic channel, an extension of the digital channel. And these do need to work in harmony. You cannot have all of the digital channels working one way and the Agentic channels work differently in terms of security entitlements, policies and so on. So what we are working on is delivering this and experience whichever way the user wants but give the control to the banks so that ChatGPT has this thing called intelligent user interfaces. So we want to work with them to continue to create -- retain the financial institutions brand experience, give the consumers the comfort that they're interacting with the right financial institution. And so that is what the value we bring as a technology partner, and we are on the hook to deliver these services, because the platform is still in the middle. The digital banking platform is in the middle where we talked about it last time. It's got the context, it's got the integration, it's got the orchestration. And this new interaction model is leveraging all of that stuff. So finally, the relationships, the data and the controls all sit within the platform. The agents are orchestrating but still are interacting with the platform in, however, the financial institution and the consumer want the agent to operate. Ultimately, the control is within these two layers.

And Adam, would you add anything to that?

Yes, I would. We talk with banks and customers and end users all the time. And I think it maybe natural to think that bringing an external agent to the digital banking experience, we tend to flatten or commoditize the experience. And I think actually the opposite is true. What we're starting to see, as Hima mentioned, is that the surface for exposing the personality of the bank, the brand of the bank, the capabilities, the MCP surface. It's wider than previous kind of mechanism. So we've dealt with IVR as a channel. We've dealt with web. Is it channel, mobile as a channel. We've dealt with the Apple Watch is a channel. We've dealt with SMS as a channel. The Agentic channel is really interesting, because it gives you a much wider surface and a broader context that you can bring to the agent. And so if you built a really boring experience, you would deliver balances and transfers and maybe bill pay. But given that the agent actually has some level of discretion and the agent is operating with an hint intent on behalf of the end user, it is possible to continue to personalize the experience and the agent still represents a proxy for the end user's intents and desires and behavior. And so I'm excited at the width of the channel that Agentic Banking produces. Because realistically, all the things Hima mentioned, entitlements, permissions, security, authentication, authorization, all of that is clearly table stakes for this space. And we will work through the Agentic protocols that come available and the evolution of that space. And we will help our banks step into an Agentic world. The upside is that the broader wider aperture for interacting with the agent and driving the agent to communicate the capabilities of the bank to the end user and the account holders through the agent and to create this richer, broader behavior. I think it's still present. And actually, it's more present than an FDX open banking interface or an Apple smartwatch interface or some of the other kind of aggregation methodologies from the past? And so I'm really excited for the opportunity that this new channel exposes because for the past 20 years, we have done a great job of helping banks absorb new pieces of technology and mapping those technology onto their channel strategy while preserving the brand of the bank, the promise of the bank and the differentiation of the bank in the market.

So when I talk to investors, I think they share a lot of the enthusiasm that we have for exploring that Agentic channel with our customers and the promise that we could ultimately help them deliver on. Simultaneously, there has been news in the headlines recently that some large banks have signaled they're not going to let outside agents connect to their institution at all on the face that could potentially sound like something that is difficult for the end market that we serve. So Adam, I'll start with you and ask, how would you want investors to think about that news? And what it means for a company like ourselves and how we serve regional community banks with our partnership?

Yes. I think to the extent that a financial institution says, I don't want to have any Agentic surface available, and I don't want agents to connect. I mean and underscores the thesis about how crucial we are to the bank and how the mobile and digital channel really drives a lot of that behavior. At equilibrium, I think this looks a little bit more like the aggregation story that started 20 years ago. Banks and credit union said, maybe I don't want to be aggregated, maybe I don't want to provide data to -- into it or later on to Venmo or to auto books or other kind of partners. And then end users they vote with their feet and their wallets, right? And they say, if it's important to them to get that connectivity, if it's important to them to get a genetic banking, they'll provide that feedback to the financial institutions and they make decisions that are reflective of the capabilities that their bank support. So I think it is rational for some set of institutions to say, "I'm not ready to have agents connect until I have a technology strategy that allows me to roll off the incremental risk of the agent workflow. " But if you think there's a bank in the United States that's going to take the position of I'm never going to allow any Agentic connections. I'm not sure that, that's durable for the long term. But if a customer in our base wanted to do that, we would absolutely support it. I think this is for now kind of decision and not a long-term decision. But there are a lot of questions to be answered around how to manage the risk in the channel to secure it effectively and to make sure that experience is aligned with the way that the bank wants to show up. And so that's where the value is. And for an FI that needs some time to do that. It totally makes sense.

Hima, would you add anything from your perspective?

Yes, maybe I'll connect what Adam said, right? It's a matter of when, not if, and it depends on when do the banks get comfortable and they're going to take their own pace at which when they get comfortable with exploring the strategy and making it. There was recently a good block from Satya, Microsoft on, what do financial institutions or any other enterprise trying to make personal agents are in a model integral part of the strategy, what are the expectations, right? Observability, auditability, control, containment, incident disclosure, and so there's a lot of these things that, again, this exposes the surface area and our financial institutions are looking for us to give that confidence and then use the technology, the digital banking and the fraud stacks to go expose the information to these personal agents, right? So that train has left as in they're using it, they're using it through experiences that are flat, like aggregators they could want richer experience, but they're waiting for us the platform providers to continue to enhance our technology strategy. And so -- and we are working with them to go continue to deliver that capabilities through the MCP endpoints that we announced a couple of weeks back at our dev days, more ways for them to go on this journey.

Great insight there. Let me shift to our next team a little bit where we talk about our risk and fraud business. And so Adam and Hima, when I talk to investors a lot this year, cybersecurity has been a pretty strong thematical trade this year, and it's not hard to see why demand story rather than a cyclical one. AI is raising the threat floor permanently and buyers are consolidating out of platforms instead of building their own defenses. And so that same AI being recognized as what these vendors sell divided, and we see this at these forces at work in our risk and fraud business. So Adam, maybe I'll start with you with how has AI changed the fraud landscape? And more specifically, where do we have the right to win relative to that ecosystem of core processors payments name and maybe even like pure-play cyber vendors that are in the space?

Yes, that's a great question, Josh. So if you'll permit me a brief metaphor. I think of the way fraud happens almost as a market clearing activity, where the threat posture of financial institutions represents the supply of available achievable targets. And then the tooling and the capabilities and the desire of threat actors to commit fraud and exfiltrate funds kind of represents demand. So AI is interesting in that it moves both curves. Because in the same way that AI makes it faster for an attacker to create a chain or complex exploit to try and drive cyber or to execute fraud with new faces or social engineering. That AI tooling is also available to the defenders on the other side to reduce the supply of available attackable targets. And so at equilibrium, I think we end up with a shift that doesn't really move the needle much in one direction or the other for institutions, that effectively take advantage of AI. And this is where our right to win is so crucial. So the first piece is, if you are deploying software that is open to the Internet and that software is authenticated, you want to be a company that does that all day, every day. And so for an FI that says maybe I can digital myself, maybe I'll vibe code it. Maybe I'll get a toolkit and I'll host it myself. The current threat landscape makes self-hosting, self-deployment, self securing and application, very, very unattractive. So that really works in our favor because we are able to more rapidly apply AI to resolve issues. Just today, there's an exploit for PHP, which runs about 70% of the public web-based Internet that everyone is going to have to go patch and address. Every one of our customers is going to call us and ask when are you guys going to have those patches in place and what's your mitigation strategy instead of turning to their internal team and saying, are you prepared for this? Can you do this? Are you ready? This is the beginning of a new form of weather and climate in the space. This is not a onetime activity. There will be more exploits. We will have to be aggressive with them. So our right to win comes from our strong operational foundation and our focus on it every day. The other piece that happens here is that the application of AI to the security workflows, application security, social engineering resistance, payment security, long-form identification of compromised accounts, there's a lot of complexity there, and it takes two things to do this well. One is the kind of talent that you have to attract and mass to a traditional technology company. And then the second is enough data to drive those models effectively. And so for our customers, the point of interdiction of the payment or the log in, it really sits with Q2. The operating orchestrating infrastructure that allows us to step into a workflow and safely secure an end user and then also let them clear payment, that maybe was identified because of a behavior that was not fraudulent, but simply anomalous. All of those things live at the orchestration layer and the system of context that we deliver. And then we have a breadth and wealth of data that a core provider or a stand-alone cyber provider doesn't have. So if you're someone that works at the perimeter and you're doing things like bot detection through key stroke cadence identification. As a fantastic technique. We have partners that do that. But they don't have the data that underlies end-user behavior inside the application end user behavior across applications or end user behavior across the financial ecosystem. So whether it's the data the talent or the operational workflow that actually require securing, the fraud position that we are in is as strong or stronger than we've ever been. And the advent of AI, even though it makes attackers and defenders both stronger creates velocity that is going to be a real tailwind for us. And so our customers are expecting and really needing us to win in this space to help keep them safe and minimize fraud losses. Without ramping friction through the roof for their customers. And that intersection is where the right to win starts for us.

That's great. Maybe you can add a little bit of color too on that and be specific around like the products and the business lines that we're attacking with that opportunity.

Absolutely. Fraud is a great business for us one of the -- it's one of the fastest-growing product lines and we talked last time about when we think about AI, the account holders experience, the back office experience, but then the fraud experience where we launched an account takeover product called [ UAM ]. It's probably one of the fastest growing products, both from how we are building it and how fast we are taking customers live. And it is all driven by the fact that as there are more channels getting in, the account takeover gets to become a bigger problem. As an example, there was account takeover works by taking signals from various data streams coming in and we didn't have a detector for applied interface, and we were quickly able to add that detector because the customers were seeing fraud coming from that channel. And so this is an example of how our fraud products are able to catch these by getting the product and the signals into the customer hands quickly. And if I simplify this from my own sake, every new channel that gets added to how customers, consumers interact when people had to walk into a branch, there was fraud but minimal because he had to depict someone else. When web came in when mobile came in, it increased because of the number of channels and the approaches you take with social fishing and all of that became bigger. And now with the Agentic channels, the operators, the fraud operators have a bigger surface area. So it becomes a much bigger opportunity for us. Because now we are looking at signals that are coming in at 20x, 30x because these personal agents are actually kicking off a bunch of sandboxes or virtual machines and using those to come into the banking financial institutions. And naturally, the fraud operators are going to go I don't know, social agent fishing, whatever the new term is going to be and start trying to use that channel. And so I look at this as even becoming more important as the surface area gets bigger for us to make that an integral part of how we think about fraud. As Adam said, now we are integrating with the cyber endpoint because that signal also becomes very important in deciding whether this is a fraud actor coming in or a legitimate action. So we are working much closer with there's endpoint, cyber. And so this merge of cyber signals coming in with the traditional, what we considered user signals coming in is happening more and more. And that enables us to do a better job, but also makes us a lot more stronger in this space. as we think about our fraud business and how fast it's growing.

Yes. A lot of good color there. I appreciate that. So maybe an adjacent conversation to that one is like security posture. So investors talked a lot about this with me and as we think about the threat environment getting more sophisticated and AI raising the stakes on both sides of the equation, both how people want to understand like a provider thinks about it like the liability that we may have serving our customers as well as our own architecture and resilience. Adam, how do you think about how important security posture is, what it means for the demand environment for our digital banking business and what customers are looking for and asking for us right now?

Yes. So security posture is really fundamental for us. What does Tom Lee Jones say in Men in Black? There's always an alien invasion. Like if you don't work in this space, you may not realize that for incident response teams and incident preparation teams, there's always a new CVE. And we've had two substantial 9-plus level CVEs this week, one around breaking out of virtual hosts into the host operating system, and that's a hyperscaler level problem that everybody is working on patching right now. And then the second is this PHP issue that's come up in the last 48 hours. And so at Q2, that cycle of identify the risk, mitigate the risk, clear the risk, make sure the risk doesn't recur. There's a team of people that doesn't really do much else. That is their day every single day, day in and day out. And turning that into business as usual is really nontrivial. What's changing is the velocity with which you have to get the patching done. And that's impacting Hima's team, to some extent about the way the [ SDLC ] is operated. And so we have a strong partnership between the operations organization that drives cloud operations. He is team on the engineering side and then the internal teams around compliance, regulatory and security. And so the more deeply they can partner and the faster they can turn the more quickly you can close whatever gaps there are in the exploit chain that attackers are seeking to pursue. This is not something you build in a month or 6 months or even a couple of years. This is something that's taken us years and years to build of operational expertise and repeated time on task among those teams. And so our customers ask us constantly about that posture for both cyber and fraud. We're refining it continuously, and we have a substantial labor investment of the -- some of the best people in the business in doing that. In addition, they are using AI and to evaluate our systems. They're using AI to improve our [ SDLC ]. And so we are working to match or potentially even exceed the velocity of attackers because posture is not static anymore. In the same way that your lead against the attacker is something that ebbs and flows as new information comes to market. And so you have to be running every day even when there isn't a new exploit or a new attack. In addition, there are new capabilities that we're bringing online. So we just recently completed a posture measurement of our agentic approach around the Assistant product. And that was done with a third-party independent auditor, the AIUC, and we've completed the [ AIUC-1 ] assessment for Q2 assistant. I'm sure that's the first of many new standards that will come about for assistant safety and sanity and quality. And so being at the front of the curve and close to market on those elements is how we establish credibility with our customers because they manage risk and they manage risk with standards and practices and controls. And so leaning into that set of capabilities and exposing to them our controls, our practices and then the way we operate is how we build trust and that trust creates stickiness for them with us, and that trust also creates the referenceability that allows us to go and win new business. So we've always talked about our investments in security posture cybersecurity operations and fraud operations, I think it's actually, again, becoming more important going forward because of the changing landscape, and we're comfortable that we can match the velocity of attackers and continue to win on that in the market.

Hima, any color from your side?

I mean I do one example reference to what Adam said as how the SDLC, the software development life cycle is sort of moving at a much, much faster pace in the context of security as more vulnerabilities are being found, the pace at which -- we have broadened security agents that look at the vulnerability, figure out the updates that need to be made to take care of that vulnerability and automating that pipeline. That's an example of how we are leveraging this technology to strengthen our posture as the PHP as an example, came out, we're able to now fix it much faster. One of the things I'm going to quote Satya, again, the most trustworthy super intelligence system will not be the one with the model we trust most. It will be the system that enables us to trust the model the least. The financial institutions are looking for a system that assumes there is more inherent risk in the technology and what is built around from a controlled security, auditability, observability, containment. And I think that is where our posture is continuing to strengthen more and more because that's how we've been working, not just about how we operate the system, but it starts from how we look at requirements. How we do our software development life cycle. How we then operate the system and how we look at our fraud SOC centers that give us feedback back into the product and the engineering teams. So it's this continuous life cycle that is core to how we deliver the system is what is getting more and more important as this technology goes forward.

Great color from you and perspective from both of you on that. I appreciate that. Let's go into something we talked about continuously this year, starting from our first session in the year with is some of the AI solutions that we actually put out into the market this year. So Hima. We're in early access, have customers across [ UAM ], Q2 assisting Q2 code. Maybe talk a little bit about what we're saying now that we put them through that cohort and where are they evaluating from these solutions?