Appian Corporation (APPN) Earnings Call Transcript & Summary

All right. Well, thanks, everybody, for joining us in the last session of the day for day 2 of the Citi Global TMT Conference. I'm Steve Enders, part of the software research team here. And with us today or for the final session today, we have Serge from Appian. Serge, thank you so much for joining us.

Thank you for having me. Great to be here.

Great. Maybe just to start off, maybe we can talk a bit about just the main use cases that Appian involves and, kind of, run through the high-level story for Appian today.

Yes. So Appian has been in business for 27 years, and what we do is we automate complex business processes. So that's a lot of words. So I'll give you some examples. We automate fraud prevention for financial institutions, mortgage or health insurance applications for health care and financial services companies, a variety of public sector use cases. One of my favorites is we run inventory of ammunition for one of the branches of the U.S. military, so that's as mission-critical as it gets. And what these use cases have in common is a few things. Number one is they tend to be cross-functional. They tend to be mission-critical and accuracy is very important. They frequently involve the customer in addition to the company itself and accuracy is exceptionally important. And the other question that usually comes with that is like, okay, well, what does Appian replace? And the answer is there's always something in place. Sometimes it's just a paper process that becomes automated for the first time. Other times, it's a homegrown application that is now having trouble scaling or simply cannot keep the functionality what the business needs or it could be a rudimentary automation tool, third-party automation tool that, again, cannot meet regulatory requirements or security requirements or accuracy requirements. So then Appian steps in and replaces it. And the answer, whatever it is, is that customers use us in order to have greater flexibility, greater agility, lower cost and in many cases, direct revenue outcomes for using us, and that's been our story for the quarter century now.

Okay. No, that's great to hear. Maybe we can talk a little bit about just the demand environment and the business environment that you're seeing right now. I think the business has been accelerating for almost a year now in every quarter. Cloud has been accelerating in the past few quarters. Just what is driving that? And how do you think about those factors and sustainability of that moving forward?

Yes. So we are seeing a very strong demand environment, and it's really driven by our AI capabilities. So Appian has been out there for multiple years now talking about AI as this powerful technology. It has tremendous benefits, but it needs a framework to actually operate at scale. We talk about this as AI in Process. And now you've heard a number of terms that have emerged for this idea, whether you talk about a harness, whether you talk about orchestration layer, it's this idea that AI needs something in order to be effective and accurate at scale. And so because we've been out there, consistently talking about AI in this context and because our product actually delivers AI capabilities and functionalities and strong returns while adhering to accuracy and performance requirement that our customers need, which are highly regulated industries in the public sector, we're seeing the demand grow. We're seeing strong pipeline that is converting, and I liken that to a wave. We have a wave of demand coming our way, and then it's our job to surf that wave or rather convert that demand into actual revenue. And we're doing a great job. Happy with execution, continuous improvements in sales productivity and you combine existence of demand with ability to close it, and that's been the story of Appian for the last couple of quarters and the story behind our acceleration that you mentioned on our cloud revenue in particular over the last couple of quarters.

Okay. I think we hear a lot about people adopting AI, people adopting agents. Can you just maybe help like crystallize a little bit more fully, like, what are, kind of, the use cases that your customers are using Appian AI for? And how do you think about the repeatability of some of those use cases to potentially spread more fully across the base?

Yes. So front and center for us right now is a use case or a functionality that we call DocCenter. So DocCenter is AI-enabled document processing at high accuracy and already a part of your process. So you're not separately processing your mortgage applications in one silo and then feeding it into some process. It's all integrated in a single process and with high accuracy and sort of ability of AI to reason and send for your next action, whatever it needs to be. The great thing about document processing is that it's a very horizontal use case. Every enterprise is drowning in documents of some sort. So it applies in every one of our verticals. It applies in every one of our geographies. And more importantly, everybody has more than one use case. So an example that comes to mind for me is a health insurance company here in the U.S. that implemented its first DocCenter use case in June. And they already need to buy incremental AI usage just to satisfy their use case, and they're talking to us about the next couple of use cases that will only further drive the need for their spending on our platform. And so I think about DocCenter as, kind of, the tip of our AI spear, and that's probably a multiyear cycle because, again, doc processing is such a pervasive use case across enterprises and certainly in our geographies -- sorry, in our industries. The next one behind it, I would say, is agents. So we GA-ed our Agentic capabilities last year, and we're seeing strong interest. It's obviously earlier on, but we're seeing, kind of, a very broad use cases from relatively straightforward simple agentic deployments to very complicated ones and very sophisticated ones and sort of across the board. So we're working this year to continue hardening the product to continue improving our ability to implement. And that's something that we're going to put more arrows and more focus on as we look into 2027. So I think DocCenter number one, agent is, kind of, like, the next and the overlapping, kind of, driver of demand. And then the third one is various products that make it easier to build applications with Appian. So Composer is the one that we talked about. We also have a functionality called DevMCP, which allows you to use your favorite vibe coding tool and use that to connect to the Appian platform and build apps using natural language on the Appian platform. So what those products will do is they will just reduce the cost of building on Appian or rebuilding on Appian, and that will be an incremental driver of growth. Obviously, that one, in my opinion, is sort of the longest in terms of the duration and very, very large, but the one that's earliest on.

Okay. Maybe on the agent piece of it, I think that one is a little bit maybe earlier in terms of the adoption curve for you all. I think when we talk to CIOs and people, it seems like agents in their view is a little bit slower for adoption. But just maybe where are we in terms of that adoption curve? And are there certain agentic use cases that are starting to pop up today for you all?

Yes. So the challenge with agentic is that really -- that true agentic really requires the right use case. And what I mean by that is that it's a case in which you need adaptive reasoning, you have a sufficient amount of ambiguity that you both want and need to deploy AI more broadly. And you still do it because it is more cost effective than applying a human, but you need to make sure that it's accurate and reliable. And again, like the agents, you want them behaving under the set of rules that you are operating. So that's sort of the sweet spot. But we see a lot of customers coming to us thinking that they have an agentic workflow, and we tell them that they're actually better off applying a more of a deterministic process with more narrow AI capabilities, which will be more accurate and less expensive. And that ends up being the better outcome for the customer. So it's not a traditional agentic use case, but it's AI-powered and it's very valuable to the customers. So again, like it comes down to ambiguity where a lot of context needs to be driven from multiple different places as opposed to 1 or 2, where incremental pieces of information are needed to, kind of, involve in step-by-step reasoning. That's where agentic is meant to be better used. And if you apply it more broadly, which some have done, not with us, but in general, you end up probably in a place where you're either spending too much money or you're not getting the accuracy that you wanted to get and then you end up going back to the drawing board.

Okay. And when people come back to -- go back to the drawing board, does that end up creating an opportunity for you to then step in and take on that use case?

Yes, absolutely. What I've generally seen is that a lot of the vendors that we run into, I feel like they've overpromised on AI early on. And this idea that AI can self-govern and just turn it over to AI and agents it will be fine. And so they're backtracking a little bit on that message. It helps our credibility that we've been pretty consistent and that we can then back up that consistency of message with the quality of our products, and that's why we're seeing the demand that we're seeing.

Okay. No, that's great to hear. Maybe, kind of, staying on this on the line of thinking, but maybe turning more towards the monetization opportunity with AI. Just how do you think about how that, kind of, develops for you moving forward? Is it more about tiering? Is it about consumption? Is it about incremental SKUs? Just how do you, kind of, think about that adoption curve and how the levers, kind of, evolve from here?

So there's multiple levers. The first one, I would say, is actually the tiering. So just to take a step back, we introduced tier pricing at the beginning of 2024, so we're 2.5 years into it. And basically, we created a standard tier, but to get access to our latest functionality and most importantly, our AI functionality, you have to upgrade to what we call the advanced tier and that runs you roughly 25% to 35% more. So just to have access to our in-production capabilities, you got to pay us out of the gate. And we talked about nearly 40% of our customers having some portion of their ARR on the advanced tier. And in Q2, we also mentioned that 85% of our new logos in the quarter actually off the street came in and bought the advanced tier, which I think speaks again to the quality of our -- credibility of our AI message. And so our advanced tier ARR has been building consistently and will continue to do so. So that's step one. Still plenty of room to go. Step 2 is usage. So our advanced tier and other ways you can buy AI come with what I would describe as a moderate amount of usage included enough for a single use case. So for example, the health care company that I talked about, their first DocCenter implementation already depleted sort of what's included. So now they need to come back and buy more AI bundles from us, more usage. And as they implement the third and the fourth use case, then obviously, all that is accretive. It's early days. Relatively few customers are at a point in which they need to buy AI usage. I think that sometimes in conferences like this, it's easy to forget that AI is still very, very early in terms of adoption really in enterprise when it comes to mission-critical applications, particularly enterprises that we serve. So it's early days, but it's building, and we think that's another medium-term lever of growth. And then the third bucket is we're going to play this game again. So above advanced tier, there's something called premium tier. Very, very few customers are in that level now. But over time, we'll start to put more advanced AI features there, and we'll ask for another 25% to 35% uplift. So we think that there's plenty of -- if you think about it both from the perspective of use cases and ability to monetize use cases, we're in early innings no matter how you think about it.

Okay. And on the premium tier, I guess, what are the capabilities that you get incrementally from -- versus the advanced tier? And I know it's still early, but like longer-term potential, where could that potentially go in terms of the penetration within the base?

So right now, it's a handful of features. Right now, it's more included AI usage. And so we are very much still in the seeding the market stage, which is what we're doing with the advanced tier. Once we feel like that game is largely behind us, which I'm not suggesting it's this year or next year. But at some point, we will start putting the incremental feature there. And then it will be sort of the next chapter of that story. But either chapters 1 and 2 have a long way to go.

Okay. That makes sense. I want to ask a maybe high-level question on AI and the opportunity set that you see here. I think we still get the conversation or the questions around like build versus buy and what that means within your -- the customer base that you serve and how they think about when it makes sense to use Appian, when does it make sense to use custom code? And how does that, kind of, view evolve over the next kind of few years? And it would be great to kind of get your perspective on what that looks like.

So thank you for saying custom code because I sometimes need to remind people that custom code has always existed. So it's not a new concept. It's not something that began with vibe coding, right? So you can always go hire yourself a bunch of developers at $250,000 a pop and have them build something custom for you. That's always been competition for platforms like ours. So nothing is really changing from the perspective of like this competitive force always existing. Okay. So now comes vibe coding or ability to use natural language coding to accelerate the process of custom coding, and it's very powerful, obviously. We've heard the customers' position change, I would say, sort of in April, May of last year. And that what we were hearing before that is, these are interesting tools. We're learning what they're good for. We would never ever ever use them for something that is very mission-critical and core to our process, but there's probably other things around that this will be an interesting alternative for. To now the conversation changing a little bit of as exciting as those tools are at the beginning, you start discovering costs associated with them down the road. And the word maintenance suddenly starts coming up. One of our partners talked to me about -- and this is a large systems integrator who built like a meaningful app using natural language processing in the hopes of demonstrating internally and externally that it can be done and then it becomes a business for them. They soon found that this app requires a couple of dozen people to be maintained. So -- and by the way, these are people who have far more technical expertise than an average enterprise. So for them to be in that situation, it shows you that this might not be tenable anytime soon for like a traditional enterprise, which is always struggling to get technical talent on board. So we see interest from customers. We see apprehension. If anything, the balance between apprehension and interest has switched, I would say, over the last 4 or 5 months more in the direction of apprehension. I'm sure they will have like a role to play, but we don't see them impacting how our customers make decisions about processes that they would consider deploying on Appian.

Okay. And at this point, it isn't having any impact in terms of like deal cycles or their decision-making to move forward with an application on Appian versus them choosing to try to build something themselves?

Not at all. Demand is strong and like deals are continuing as they have been.

And pipeline looks good.

Pipeline is very good.

All right. That's great to hear. Maybe shifting gears a little bit with this being a big Fed quarter, and I think that's about 1/3 of the business, something like that for you all.

Federal government, 25%, total government, low 30s, yes.

Okay. Just maybe we can talk a little bit about the demand trends that you're seeing within that vertical. I know that there's been a lot of automation initiatives with DOGE over the past couple of years. There's a massive $500 million ELA there as well. Just how do you kind of view that opportunity moving forward? And any kind of views for what that means for this quarter and kind of like what you're assuming in the guide at this time?

Yes. So there's been a positive structural change in the federal sector for Appian last year. And what I mean by that is the government is focused on efficiency. The government is focused on impact of its technology initiatives. Ongoing never-ending projects are out of favor and dealing directly with vendors who are going to sell you software and implement that software so you can deliver flexibility, agility, cost savings, whatever it may be, to your use case is what the government wants. And that's simply put us in a better competitive position than we've been before. We are more frequently directly competing as opposed to resellers. There's greater focus on sort of the value of the software as opposed to just its price. And that speaks to the strength -- that speaks to our strengths. We've had a very strong year across the board in federal last year, including a strong close in Q3. And frankly, to me, the question was, is this a onetime thing, meaning the first year of the new administration? Or is this like an ongoing change? And sitting here 6 to 9 months later, I can tell you that it feels like an ongoing change. So that's why it's like not a onetime thing, but a secular demand driver. We expect to have a very strong quarter in Q3 in federal. That's part of our guidance for the year. And we expect to have a strong year in federal next year.

Okay. And I guess the Army ELA that you have, how do you kind of view, what that means and I guess, the repeatability of that with other agencies within the federal government at this time?