APPN - NASDAQ

Good day, and thank you for standing by. Welcome to the Appian Fourth Quarter 2025 Earnings Conference Call. [Operator Instructions] Please be advised that today's conference is being recorded. I would now like to hand the conference over to your first speaker today, Brian Denyeau from ICR. Please go ahead.

Good morning, and thank you for joining us. Today, we'll review Appian's fourth quarter 2025 financial results. With me are Matt Calkins, Chairman and Chief Executive Officer; and Serge Tanjga, Chief Financial Officer. After prepared remarks, we'll open the call for questions. During this call, we may make statements related to our business that are considered forward-looking. These include comments related to our financial results, trends and guidance for the first quarter and full year 2026, the benefits of our platform, industry and market trends, our go-to-market and growth strategy, our market opportunity and ability to expand our leadership position, our ability to maintain and upsell existing customers and our ability to acquire new customers. These statements reflect our views only as of today and don't represent our views as of any subsequent date. We won't update these statements as a result of new information unless required by law. Actual results may differ materially from expectations due to the risks and uncertainties described in our SEC filings. Additionally, non-GAAP financial measures will be discussed on this conference call. Reconciliations of GAAP to non-GAAP financial measures are provided in our earnings release. With that, I'd like to turn the call over to our CEO, Matt Calkins. Matt?

Thanks, Brian. Thanks, everyone, for joining us today. In the fourth quarter of 2025, Appian's cloud subscriptions revenue grew 18% to $117.0 million. Subscriptions revenue grew 19% to $162.3 million. Total revenue grew 22% to $202.9 million. Adjusted EBITDA was $19.7 million. For the full year, Appian's cloud subscriptions revenue grew 19% to $437.4 million. Subscriptions revenue grew 18% to $576.5 million. Total revenue grew 18% to $726.9 million, adjusted EBITDA was $76.8 million. 2025 was a successful year for Appian for several reasons. First, we executed our strategy to sell big deals to leading organizations. The number of customers that purchased over $1 million of software this year grew 50%, nearly doubling the value of our 7-figure transactions. I'll share 2 quick examples. A European pharmaceutical research organization purchased a 7-figure software deal to digitize its clinical trial site selection. Appian will accelerate its selection process with AI, improving patient selection efficiency and reducing trial costs. Separately, a North American aerospace manufacturer purchased a 7-figure software deal to automate a core manufacturing system and save the company nearly $60 million over the next 3 years. The second reason Appian had a successful 2025 is that our position within the U.S. public sector strengthened partly due to structural changes. We closed big deals as the new administration have emphasized efficiency and changed how it purchases and implements technology. For example, a U.S. military branch named Appian its cornerstone platform to modernize operations and increase efficiency. In Q4, it signed a 7-figure software deal to unify systems and deploy them to over 100,000 users. The federal government has shifted to partner more directly with software vendors and reduce its reliance on intermediaries. Appian stands to benefit as indicated by the enterprise agreement that the U.S. Army awarded us this quarter. The Army is already an 8-figure ARR customer. This new framework allows it to purchase $500 million in Appian software and services over the next 10 years. This agreement shows the Army's ambition and commitment to use Appian to modernize systems and transform operations with process and AI. The third reason Appian had a successful year is that we continue to increase our operational efficiency. We've now increased our go-to-market efficiency in 10 sequential quarters. You know this metric means a lot to me, I always mention it. Appian generated 11% adjusted EBITDA margin for the full year 2025 compared to just to negative 8% just 2 years before. We created $63 million in operating cash flow compared to a loss of $110 million 2 years ago. Credit these efficiency improvements to tighter resource allocation and sales, global diversification and back-office AI enhancements. We're creating an operating model that's built to drive further margin expansion going forward. Appian's strong financial performance puts us in a position to start consistently returning capital to shareholders. Today, we're announcing a $50 million stock buyback. Finally, I'll tell you the best thing about 2025. The best thing is that it's become common knowledge over the past 6 months that AI needs process, also known as workflow. Without a process framework, AI cannot add value to complex work streams or collaborations. Market analysts and researchers like Gartner and MIT published papers on the topic. Customers, prospects and partners have all confirmed the trend. Our competitors shifted their messaging, began talking about workflow and added rudimentary process technology. This trend validates Appian's long-standing position on the issue and recognizes the synergy between AI and process that we built our strategy around. I'll take a moment to explain why AI needs process. AI is probabilistic, which is to say it's slightly unpredictable. Most important work at the large organizations Appian targets requires total reliability. So AI needs a deterministic framework like our process layer. That deterministic layer provides direction and guardrails and certain functionality you wouldn't ask AI to write on its own. Code is becoming cheap, but mistakes aren't. So the more important to the work, the more essential is the deterministic layer. In the coming years, AI will do a lot of work and write a lot of software, but it's not going to do it alone. Where AI goes, process must also go. Our technology is an essential enabler of AI. Appian has been a process leader for more than 20 years. We were a pioneer in this market back when they called it business process management. We led providing BPM in the cloud. We led in building a process-centric suite. Our unitary platform provides workflows, data fabric, process mining and built-in security. We led again embedding AI in our processes. We've earned trust at the largest firms and are executing mission-critical processes to the highest standards. 2/3 of the world's top 10 life science firms, asset managers and non-Chinese banks are Appian customers as well as all 15 cabinet-level agencies and military branches in the U.S. government. These groups use our platform for complex mission-critical processes like customer onboarding, claims management, patient intake, regulatory compliance and procurement. Exponential growth in our AI traffic shows that our platform is becoming an AI vehicle for large organizations. AI use on our platform grew 14x year-over-year. AI use on our platform grew 14x year-over-year, and we are monetizing that growth. Customers must upgrade to Appian's AI license tier, which comes with an average price increase of 25%. A meaningful number of customers make this upgrade every quarter, including this quarter. Much of our revenue profit and pipeline growth in 2025 is a result of our synergy with AI. Most of the 7-figure software deals we booked this year were driven by a desire to access our advanced features like AI. Here are 3 examples. First, a leading pharmaceutical company deployed Appian AI into an existing application this quarter. The application tracks interactions between the company's sales team and health care practitioners to ensure compliance with international regulations. We're deploying an Appian product called Doc Center that uses AI to parse incoming e-mails, documents and other communications. Doc Center uploads data, pre-populates forms, triggers workflows and accelerates response times, in this case, by 88%. Next, a top advocacy organization representing over 100 million Americans and 7-figure ARR customer, named Appian an enterprise standard this year. It recognized the importance of deploying AI within an Appian process after evaluating various AI vendors. In Q4, it purchased a large upgrade to access our latest AI features. The group will deploy Appian AI agents to reconcile tens of thousands of invoice payments annually. Reconciliations used to take over an hour per invoice, now the organization expects to complete tie-outs in just minutes. Finally, a network of European banks signed a 7-figure software deal this quarter to access our latest AI features. The group already runs know your customer and loan overdraft processes on Appian. In Q4, they named our platform as an enterprise standard for modernizing core processes. The conglomerate will use Appian Doc Center to classify and extract data from dozens of documents to open cases for processing. The banks expect to save more than EUR 20 million over 3 years as they scale operations. Recent market moves show investors are concerned that AI poses an existential threat to software firms, including Appian. There are 2 main worries. First, the AI will do all the work that software used to do; and second, that AI will write all the applications. I'll address each point. First, Appian leads in the technology that AI cannot thrive without. It's becoming understood now just how much AI needs process. AI is probabilistic technology, not reliable enough for the highest value use cases. Unpredictability is an indelible part of AI's identity. Years of improvement will not make it otherwise, nor will enterprises ever decide to accept AI level unreliability. A deterministic layer is essential. Something to direct the work, something to detect and remediate the errors, something that can produce perfect outputs from imperfect efforts. Process and workflow is that technology. Long before AI, process orchestration was developed to best utilize that other unpredictable worker, the human being. As repeated studies attest, AI is not yet transformative in the enterprise. In PwC research last month, most CEOs report AI having no impact on revenue or cost. But impact is coming when AI is connected to valuable work streams, and Appian is leading the way. With the help of a process layer, AI will be a very productive worker indeed and very widely deployed, providing a framework so that AI can address the world's most important work. It's like selling pickaxes in a gold rush. Second, about AI writing code. We sell to our customers value and safety, not code. Approximately 80% of our revenue comes from highly-regulated industries and the government sector. Customers buy Appian for performance, precision and peace of mind. We sell compliance to regulations, reliable customer service and accurate decisions. We sell the reassurance of a community of practitioners and 24-hour expert support. AI-generated code cannot provide these things. Only with Appian's deterministic framework, can AI create applications and perform work to meet the most exacting requirements. This is the reason why no Appian buyer has ever suggested to me that they would vibe code a critical system. They know better. This current concern about AI-generated code reminds me of the open source scare years ago. Open source seemed to threaten the pricing power of the entire sector. But in the end, it proved that code isn't the center of value in enterprise software. The value comes from the community and the support and the corporate commitment to reliability. Since open source became a popular term in the late '90s, the global software industry has grown by a factor of over 5x. Appian has faced open source competitors in our market. They appealed best to low-end buyers and had no impact on our growth. I expect AI-generated code to be adopted in mistake tolerant and low-value use cases. To write enterprise code, AI needs a platform like ours that facilitates careful specification, developer collaboration, revision and the strategic reuse of preexisting assets. In conclusion, the more organizations use AI, the more they need process orchestration. Process mitigates AI's shortcomings. Together, AI and process can address the world's most critical jobs, but AI cannot do it alone. Before I end my segment, I'd like to welcome Dave Link to Appian's Board of Directors. Dave is an expert in scaling enterprise software companies and applying AI to complex globally distributed systems. He is the CEO of ScienceLogic, an AI-driven observability and IT operations platform. I'm excited to welcome him to our team. I also want to thank Jack Biddle for his exceptional contributions over the course of many years on the Appian Board. And with that, I'll hand the call to Serge.

Thanks, Matt. Before turning to our fourth quarter results, I want to cover some changes that we are making in our reporting in order to give investors better insights into our financial performance. First, we have reclassified certain IT, cybersecurity and facility expenses from our G&A expense line item into other line items in our P&L. There is no change to our total expenses, just which line item they are shown in. We believe this new presentation of our financial is more comparable to those of other software companies. Second, we are introducing a new metric, cloud net ARR expansion. This metric is calculated by taking the ARR of our cloud customers at the end of the prior year period and measures the ARR of those same customers at the end of the current quarter. We report cloud net ARR expansion in constant currency. We believe this metric gives investors a more timely insight into our business and is more comparable to how other software companies report expansion from existing customers. Going forward, we will no longer report cloud gross renewal rate and net revenue retention. Finally, we refine our definition of a customer. We now aggregate entities based on their ultimate parent company or an equivalent government entity, whereas previously we counted at a more granular level. As with our other changes, we believe this new methodology is more common practice. Please refer to the earnings call supplemental deck for further information on these changes. Now let me turn to our Q4 results. We had a strong quarter of new business driven by continued AI traction and ongoing momentum in our focus on the high end of the market. The standout performer was our commercial North America theater with the fastest new business growth in over 3 years. Cloud net new ACV bookings were approximately 76% of total net new software bookings in Q4 compared to 65% in the prior year. Q4 cloud net new ACV growth was the strongest we've seen in almost 3 years. Appian met or exceeded the guidance ranges we provided on our key metrics of cloud revenue, total revenue and adjusted EBITDA. Cloud subscription revenue was $117 million, an increase of 18% year-over-year. We achieved the high end of our guidance even as FX contributed approximately $1 million less than what was assumed in our guidance. On a constant currency basis, cloud subscription revenue increased 16% year-over-year. This quarter was more back-end loaded than normal in terms of new business, resulting in relatively little revenue contribution from new business in the quarter. Our constant currency cloud ARR growth, which represents the exit run rate was stable versus Q3. Total subscription revenue was $162.3 million, an increase of 19% year-over-year. On a constant currency basis, total subscription revenue grew 16% year-over-year. Professional services revenue was $40.6 million, up 36% compared to the fourth quarter of 2024. Total revenue was $202.9 million, an increase of 22% year-over-year. On a constant currency basis, total revenue grew 19% year-over-year. Our cloud net ARR expansion was 114% in Q4 compared to 113% a year ago and 112% in the prior quarter. The uptick was driven by a particularly strong quarter of upsells to existing customers in Q4. We ended the year with 140 customers with $1 million plus of ARR compared to 115 a year ago. Now let's turn to profitability. Non-GAAP gross margin was 73% compared to 77% from the year-ago period and 74% in the prior quarter. Our subscription non-GAAP gross profit margin was 86% compared to 88% in the year ago period and 86% in the prior quarter. Professional services non-GAAP gross margin was 23% compared to 27% in the year ago period and 31% in the prior quarter. Total non-GAAP operating expenses were $131.5 million, up from $109.8 million in the year ago period. Adjusted EBITDA was $19.7 million, ahead of our guidance of $10 million to $13 million and compared to adjusted EBITDA of $21.2 million in the year ago period. This outperformance relative to our guide was largely driven by greater-than-expected revenue. Non-GAAP net income was $11.1 million or $0.15 per diluted share compared to a non-GAAP net income of $13.2 million or $0.18 per diluted share for the fourth quarter of 2024. This is based on 74.9 million diluted shares outstanding for the fourth quarter of 2025 and 74.6 million diluted shares outstanding for the fourth quarter of 2024. Turning to our balance sheet. As of December 31, 2025, cash and cash equivalents and investments were $187.2 million compared to $159.9 million at the end of last year. For the fourth quarter, cash provided by operations was $1.1 million compared to $13.9 million for the same period last year. For the full year 2025, cash provided by operations was $62.9 million compared to $6.9 million in 2024. Turning to guidance. We are expecting to deliver another year of solid cloud subscription revenue growth and our third consecutive year of adjusted EBITDA margin expansion. Our focus is on consistent execution and capitalizing on the opportunity in front of us. Starting with the first quarter of 2026. Cloud subscription revenue is expected to be between $119 million and $121 million, representing year-over-year growth of 20% at the midpoint of the range. Total revenue is expected to be between $189 million and $193 million, representing year-over-year growth of 15% at the midpoint. Adjusted EBITDA for the first quarter of 2026 is expected to be between $19 million and $22 million. Non-GAAP earnings per share is expected to be between $0.16 and $0.20. This assumes 75.1 million fully diluted weighted average shares outstanding. For the full year 2026, our cloud subscription revenue is expected to be between $502 million and $510 million, representing year-over-year growth of 16% at the midpoint of the range. Total revenue is expected to be between $801 million and $817 million, representing year-over-year growth of 11% at the midpoint. Adjusted EBITDA is expected to range between $89 million and $99 million for an approximately 12% margin at the midpoint of the range. Non-GAAP earnings per share is expected to be between $0.82 and $0.96 or approximately 46% growth at the midpoint. This assumes 74.8 million fully diluted weighted average shares outstanding. Our guidance assumes the following. First, we anticipate our non-cloud subscription revenue to be roughly flat on a year-over-year basis in Q1 and in 2026 as our customers are increasingly opting for the cloud. Second, we expect professional services to grow in the teens in Q1 and high single digits for the full year. Third, total other income and interest expense will be approximately $3 million in Q1 and $12 million for the full year 2026. Fourth, our guidance assumes FX rates as of mid-February. Please note that we expect FX benefit to our reported revenue growth rates in Q1, but we expect FX to be roughly neutral to year-over-year growth for the rest of the year as we annualize the U.S. dollar depreciation from April of last year. Finally, as discussed previously, after 2 years of relatively flat OpEx, we are returning to a moderate pace of investment in 2026. We are investing in the growth of our sales org as well as the expansion of our engineering capacity in India. Despite these investments, we are forecasting 1 percentage point of adjusted EBITDA margin expansion in 2026. Before wrapping, let me also touch on our share repurchase announcement. As most of you know, we are very careful about dilution as evidenced by our stock-based compensation expense as a percent of revenue, which is less than half that of other software companies our size. As Matt mentioned, thanks to significant improvement in profitability over the last 2 years and becoming a meaningful cash flow generator, we are in a position to announce a $50 million share buyback. We expect this program will essentially offset the dilution from stock grants issued this year. We see this buyback authorization at the beginning of a consistent capital return policy for our shareholders. Our intention is to scale the size of our share repurchase program in line with the growth in our cash flow in the coming years. We will look to execute on this buyback during 2026. In closing, we are pleased with our Q4 results, in particular, our traction with AI and believe we are well positioned to deliver a successful 2026. We are excited about the opportunity ahead, and we'll continue to invest responsibly to maximize our long-term value. Before we move to Q&A, I'd like to invite you to our Investor Day in New York on May 14. We'll be sharing updates on our product and strategy, and you'll have the opportunity to hear directly from our customers. If you'd like to attend, please reach out to [email protected]. Now we'll turn the call over for questions. Operator?

[Operator Instructions] And our first question comes from the line of Sanjit Singh of Morgan Stanley.

This is Oscar Saavedra on for Sanjit. Yes. Congrats on the great quarter, guys. Nice to see the cloud net expansion uptick quarter-over-quarter. I was thinking maybe on the guide, it looks like Q1 guide is a bit of an acceleration from Q4. I imagine part of that is that expansion ticking up. But maybe can you help us understand a bit more the visibility and the confidence that you have given that acceleration?

Yes. So we're happy with how we wrapped up 2025, and we'll set up well for 2026 as evidenced by the full year growth rate of 16% for the year. Q1, I guess, 2 things I would say about it. Number one, it will benefit from strong new business that we had in Q4, which is why on a sequential basis is a robust guide. And then the other thing that I would call out is just that it is a quarter in which we'll still benefit from a meaningful FX tailwind. And that's clearly the primary difference between the full year guide and the Q1 guide.

And our next question comes from the line of Raimo Lenschow of Barclays.

Perfect. Congrats. That was a great Q4. I have 2 questions, one for Matt, one for Serge. Matt, if you think about -- I'm totally aligned with you with your vision around AI and agentic needed like a control layer. Like how do you think what gives you the right to be that? Because like, obviously, a lot of other people are kind of trying to eye for that because that kind of position will be very strategic as well. So talk a little bit about what Appian brings to the table that you can go to customers and say, like, I should be that layer. And then I have one follow-up for Serge.

Yes. Well, I want to say that we've been that layer for a long time. We've been that layer before large language models exploded onto the scene. We've been embedding AI actors, you call them agents, in our software, in our processes for about a decade doing jobs as digital workers because we've been a platform that enables and governs digital workers. So we're not a Johnny-come-lately to the idea of governing a digital worker or an AI agent. In fact, it's been our business, and we've been leading for a decade. So I think it's a natural for us to inherit this position as well. But I also want to say that because we have such a strong governance layer, such a unique ability to detect and remediate errors, such a monitoring layer and a self-improvement and an optimization layer, I think we're really uniquely equipped for this moment. I think there's some other vendors that went all in on agents and then realized they needed a governing layer, whereas we come into this market with a governing layer and are, therefore, really well equipped to give agents the structure they need in order to succeed.

Okay. Perfect. And then, Serge, if you think about the slight increase in OpEx you talked about on the sales org, et cetera, how do you think about the evolution of sales capacity from here onwards? And I'm asking because it does feel like a whole new world is opening, and there's quite a few players in the software space that are now thinking about like we probably should think about sales capacity increases. Is this just one-off things? Or how do you think about that evolution here?

Thanks, Raimo. So I guess I'll start with a little bit of history. We've done a great job significantly improving our sales productivity and paybacks on our sales and marketing investment, really particularly last year. And that, frankly, gives us the right to grow our sales org because we want to do it in a financially responsible way. So that's point number one. Point number two is kind of like your point, which is the market is large and growing, and we are very underpenetrated versus the opportunity. So to us, this return to growth of the sales org is the beginning of a long-term trend. But it's important to do it consistently over time. What you don't want to do is overextend because it's a difficult operational task. What you want to do is bring in people, make sure they're successful, make sure that they reach their productivity and then do it again year after year. And that's fundamentally how you kind of put yourself in a position for multiyear growth.

Our next question comes from the line of Steve Enders of Citi.

Okay. Great. I guess I just want to start on maybe the opportunity around AI. And I guess, given the purview that you have in some of these larger customers, just what have you seen so far from how their budgets or their purchase decisions are changing as they're looking to incorporate AI? And I guess maybe what does that mean? Or I guess, how do you kind of view the opportunity pipeline and given what that means for '26?

Yes. AI has been an unalloyed positive for us in our relations with our customers, maybe causing some consternation in the investing market. But in our sales situation, it's an entire positive. It gets us into higher-level conversations. It allows us to speak strategically to the top topic that's on executives' minds. We are more likely to win according to internal analyses when AI is a factor in the decision. So it's helped our TAM. It's helped our access. It's helped our win rate. We're benefiting in all dimensions from AI.

And Steve, if I can just add, just to give you a sense of how that's sort of playing out over time. Customers begin with proof of concepts. Then when they are ready for production use case, they need to upgrade to our advanced tier to have access to AI and production. And we talked about in the past about how the percentage of our customers that is on that tier is growing. And -- but plenty more that we can upgrade to advanced tier. Then what we're starting to see is some of those customers have already upgraded coming for the second or the third workload because they're happy with the performance of the first one. And that gives us incremental opportunities to grow revenue there. And then over time, we will have incremental tiers as more functionality comes online. So that's kind of the process of upselling AI and how it fits into a company's budget.

Yes. Let me follow on that. We've got this thesis, and you've heard it because I just talked about it, that AI belongs in a process and that within the deterministic framework of process orchestration, AI can really attach itself to valuable work and create new value. So we've detected that some of our solutions are ideal vehicles for demonstrating that thesis. And I mentioned in my prepared remarks, this solution called Doc Center, which is a pretty straightforward usage of our technology. Just ingest documents, launches workflows, uploads data, rapid turnaround, high accuracy. But it's just such a good demonstration that we are focused on driving this into dozens or scores of accounts as quickly as possible this year as we can because everybody who sees this knows our thesis is correct. And this is what I think we need to do in the market. We need to establish that our philosophy of making value out of AI is accurate. Every organization in the world right now is wondering how they can make use of AI. We're wondering whether AI can have a value proportional to its CapEx, and we have on our hands a demonstration of how to make that value. Just embed it within a process and it goes. And the results are tremendous and it's predictable and we can install it quickly. So where we see that we've got some of these winning demonstrations that establish how you could make value with AI, we're going to put the pedal down.

Okay. That's great to hear, and I appreciate the context there. Maybe on the Army enterprise agreement. I appreciate the color on the 8-figure customer there. But I guess kind of where do you see that spend potentially going? Or how do you kind of view, I guess, what incremental use cases or how you just kind of view that relationship developing moving forward with the enterprise agreement?

Yes. This is a threshold for us. This is an important moment in the growth of this organization. It represents a degree of confidence that an agency has not in the past shown in us. We've done a lot of great work, and we've done some big projects and delivered some wins, but we've never had a $500 million ELA like we do now with the Army. And that speaks volumes inside the Army. It allows us to speak to any part of the organization with great credibility, but it also allows us to go to other departments in the government and say, here's the department that knows us best. Here's evidence that -- of what they see in us. It also allows us to approach our partners and say like this is the kind of -- this is what we could succeed on together, and now we want you to help us somewhere else. So this is just a wonderful badge of seriousness, and we're going to wear it all around Washington. Really pleased with what that says. As for how we got it, I'd say a lot of the conversation is around modernizing legacy applications. And this is something I've spoken about on previous earnings calls, but I didn't get into it much this time. But I do want to mention that this topic is causing a lot of excitement amongst our customers and prospects. If I mentioned or demonstrate even better the technology that we have today to convert a legacy application into a modern Appian application, it typically stops the conversation cold. No matter what it is we were talking about, if there's a customer or prospect executive in the room, they want to stop everything and talk about legacy modernization. And we had conversations on that at the Army, who obviously has a number of legacy applications of their own, and that provided a lot of the momentum behind this award.

Our next question comes from the line of Derrick Wood of TD Cowen.

Matt, I appreciate the thoughts on the landscape of AI versus software, given all the concerns out there. My question is when it comes to building software applications and processes for your customers, how are you guys using AI internally to accelerate that value delivery? And then when it comes to the LLM vendors, like what do you think the challenges they might have in trying to build their own software orchestration and governance layer up the stack?

Yes. Okay. So they are going to have some challenges, and it makes a world of sense for them to partner with us in order to complement the power of their model. Look, the whole industry is under pressure right now. In 2026, it's a time of testing, where AI has to demonstrate that it can create commensurate value to justify the CapEx. And I think it's the question on everybody's lips and every organization is wondering about it and the 56% of CEOs who reported no value in the PwC survey last month are wondering about it. Everyone is wondering where we will find the value. And of course, the answer is actually very simple. You just have to connect AI to the processes where the greatest value takes place. And that's a slightly complicated connection in order to pull off because AI needs role and responsibility and a constrained aperture functioning and checkups and revisions and learning and so on. It's just -- it needs what a process layer could have given it. And so I feel like this top question that looms over the economy in 2026 is a question that I won't say we have the answer to it, but I say we have a lot to say. We have a lot to say about this question. I'm excited about the ability to prove that answer in concert with the large language models. Of course, we're agnostic. We work with all and many of them. And for that matter, with the clients who are all desperate to find an answer to this question. They're all eager for value, but not wanting to take a risk and to move first and to run afoul of the many pitfalls in AI, the unreliability and the dangers that come with that. What was the first part of your question again?

Just how you're using AI internally to maybe help accelerate the value you deliver to customers?

Yes, that's right. Well, we're using it thoroughly, right? We're expecting major increases in all of our development capabilities this year because we're making a prolific use of AI. So I expect that to be excellent for our productivity and engineering. Also, we're using it in every deployment. So when our services teams are on site, creating new applications for our customers. They are invariably using AI, which is terrific for acceleration for optimization, for recommendations on improvements, just a marvelous way to get to the endpoint. And I want to clarify here that the endpoint is not a stack of AI written code. The endpoint is an Appian application, which provides the structure, the guardrails, the safety, the monitoring that AI alone wouldn't have provided. And also that Appian application, once you've used AI as the bridge to an Appian application, that application now has the flexibility, the ability to evolve over time to match new strategic needs or to cultivate greater efficiency or to leverage new technologies. It is a living vehicle instead of what you could call new legacy, right? You don't want to go from old legacy to new legacy. You want to move to a living vehicle that can adapt as your business evolves.

Great. And then for Serge, I mean, you guys had 36% growth in professional services, I think that was the highest in 8 years. Your on-prem business was quite strong as well. It doesn't sound like you expect that to continue in the upcoming year. Could you just give a little more color on what drove that outsized strength that seems to be a little more onetime?

Yes. Let me take them in order because they're different answers. So we've been very pleased with the demand we're seeing in our professional services business for -- especially in the back half of 2025. And it comes down to a couple of things. One is the world of AI because as Matt was just talking about, customers want to get the value but they're sensitive to get it at the levels of accuracy and performance that they are accustomed to. So when they choose our software, they usually also partner with us on implementation. Because we've done it before, we bring that implementation know-how, which is scarce in the market right now. And that helps us kind of sell both software and services when it comes to AI. And then the second piece is federal. Our success there and the change in how the government likes to deal with vendors has helped our professional services business on the federal side as well. And that really drove our business next year. And frankly, we're expecting it to continue driving that business next year with 9% growth rate. We did see an uplift in demand, and we're going to continue seeing growth there, but it's not going to be a step function as we have experienced here in the back half of 2025. And then on the on-prem side, we had a very strong Q4. Frankly, that was all federal. We credit to our teams when the shutdown ended, we were ready to go and we got the deals that we were going to get, frankly, even better than we would have expected had the shutdown not been there. So that's the story of the fourth quarter. But then as you look forward, I can tell you what we see quantitatively and qualitatively. On the quantitative side, we just see a bigger mix of cloud in the pipeline than has been the case historically. And then secondly, when we talk to our customers, even our on-prem customers, they are looking for incremental deployments in the cloud. So for example, one of the largest deals that we have in the pipeline in Q1, we'll see if we get it or not, is for a customer who did one of our largest on-prem deals last year. And that's not them moving their Appian workloads from on-prem to the cloud. That's incremental deployment in line with their own IT strategy and moving to the cloud, which is why the description or the forecast for the on-prem business is what it is.

Great. Congrats.

Our next question comes from the line of Devin Au of KeyBanc Capital Markets.

All right. First one I have, maybe for Serge. Could you maybe speak to the framework of kind of the '26 revenue guidance. I believe last year, given some leadership transition, some uncertainty around pub sec and changes around go-to-market, there could be some more conservatism being embedded in the initial guidance '25. Are you applying kind of similar framework here in '26? Or can you just speak to that a little bit more?

Yes. So how we forecast the business hasn't really changed internally, and there's no incremental conservatism or caution. From a macro environment, I think I can speak for the company, even though I wasn't here a year ago. It feels a lot less uncertain than it did a year ago back when [ DOGE ] was starting, and there was a lot of macro sort of headwinds or potential headwinds related to the international relations. So from that perspective, we feel like we have perhaps a better handle on the world out there than we did a year ago. The thing that I would say specifically, though, is I divide the guide into cloud and the rest of the business. As you can see, the cloud, it's just ratable. It's frankly a little bit easier to forecast, which is why the range there is narrower for the full year as a percent of total business, whereas then we have a broader range as a percent of the business for the rest of it just because as you've even seen last year, for different reasons, both on-prem and professional services can be lumpier. And that's why the range on the full year guide is as wide as it is.

Got it. I appreciate the context there. And then just a quick follow-up on the strength that we're seeing from pub sec. You continue to see momentum there, which is encouraging, and it seems like Appian is really well positioned there. As you guys kind of return to sales capacity growth, could you just speak to like how are you thinking about the deployment of resources towards that vertical specifically and kind of how you guys are going to sustain and amplify the success there?

Yes. We are growing our capacity in the federal vertical, but we're growing it in other verticals as well. At the end of the day, I will just reiterate what I said, the size of our distribution is a limiting factor versus the size of the opportunity. And we don't want to try to address that in a big bang because then you run the risk of deteriorating execution. So we're going to hurry up slowly, and we're going to build sales capacity year in and year out. But certainly, that's the case in the federal space as well.

Congrats on the strong results.

And our next question comes from the line of Lucky Schreiner of D.A. Davidson.

Great. I'll echo my congrats as well. I have a follow-up question on the guidance. Coming off a strong quarter, the cloud growth guide, there's a lot of deceleration baked into that throughout the year. So just I'm wondering, is that all FX related? The pipeline sounds strong. So is there maybe conservatism around deal timing or ramping of sales capacity? Just curious what's driving your outlook on specifically the cloud growth guide.

Yes. So cloud growth is 20% at the midpoint for Q1 and 16% for the year. And the majority of that really isn't anything about the underlying constant currency business. It's really about the fact that we still get one more FX bump in Q1 before it normalizes.