This call is being recorded. Investor relations at Five9.
Thank you, operator. Good afternoon, and welcome to Five9's third quarter 2026 earnings conference call. With me today are Amit Mathradas, Chief Executive Officer, and Bryan Lee, Chief Financial Officer. Our discussion today will include forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995, including statements regarding our quarterly and full year 2026 guidance, business and industry trends, and other future events or results. These forward-looking statements are simply beliefs and predictions. We caution you against placing undue reliance on these forward-looking statements, and they are subject to substantial risks and uncertainties that could cause actual results to differ materially from the statements being made today. We encourage you to review our most recent reports, including the risk discussions under the caption Risk Factors and elsewhere in Five9's most recent 10-K and 10-Q.
We are not obligating ourselves to revise our results or any of the forward-looking statements we make today. In addition, management will refer to non-GAAP financial measures during this call. A discussion of why we use non-GAAP financial measures and a reconciliation of our GAAP versus non-GAAP results and guidance is currently available in our press release issued earlier this afternoon, as well as in the appendix of our investor relations deck that can be found in the investor relations section of Five9's website at investors.five9.com. Also, please note the information provided on this call speaks only to management's views as of today and may no longer be accurate at the time of a replay. Lastly, a reminder, unless otherwise indicated, financial figures discussed are non-GAAP. Now, I'd like to turn the call over to Five9's CEO. Please go ahead, Amit.
Thank you, Tony, and good afternoon, everyone. Q3 was another solid quarter for Five9. We delivered revenue above the high end of our guidance at $323 million, with subscription revenue growth accelerating to 16% year-on-year. Our CCaaS revenue grew 8%, and AI revenue grew 91%. Over the last two earnings calls, I've laid out four priorities for Five9: building a performance-driven culture, improving our operations, strengthening our core business, and winning an AI-powered customer experience. All four remain relevant today, and we are making steady progress on all of them. This afternoon, I am going to go deeper into two of them, our CCaaS platform and AI. We believe Five9's role in customer experience is becoming increasingly important. It starts with our CCaaS platform and the technology that connects customers to the person who can help.
The contact center is where customer experience is won or lost, and our deep understanding of customer workflows and routing capabilities determine how every customer interaction gets to the right outcome. That position gives us a clear advantage over companies that do not own the incoming voice pipes or have core routing and telco capabilities. Our view is that winners in customer experience will not be companies that automate the most. Rather, they will be the ones that automate the right things. Gartner expects agentic AI to resolve 80% of common service issues without a human by 2029. To us, the right question is, which 80%? Humantic CX is our answer. Automate the work that AI can resolve and equip people to handle the situations that need judgment and empathy, all while making the transitions between AI and humans work seamlessly.
When people handle the important interactions, reliable voice, routing, and tools built for human agents remain absolutely essential. A customer needs to reach the right person, and that person needs the information and tools to resolve the issue. We have spent years building and operating the foundation in these types of complex enterprise environments. Now, we are using AI to make the foundation even more efficient. Our new products, like Genius Routing, match the customer needs with the skills and proficiency of the available agent. Other Five9 AI products help agents prepare for the conversation, find information, and complete the work afterward. Applied across the interaction lifecycle, AI lets human agents spend more time helping customers and resolve cases. The right balance between humans and AI depends on what the customer is trying to do. Someone checking an order or resetting a password may simply want the fastest answer.
Simple requests like that are well-suited for AI. On the other hand, someone dealing with a denied insurance claim is more likely to want a person who can explain the decision and take responsibility for what happens next. Even as AI improves, we firmly believe that people will continue to value human attention in those high-stakes situations. Our focus is therefore where we have the strongest fit: financial services, healthcare, insurance, and other complex or regulated environments where human judgment and accountability will remain essential. In these settings, AI handles routine requests, retrieves information, takes permitted actions. It can also use hold time to run checks and gather context so the human agent arrives fully informed. People handle sensitive decisions, exceptions, disputes, and situations where empathy or accountability changes the result. What makes this work well is the seamless handoff. Five9 connects the workflow and carries the customer's context forward.
The handoff works when AI and the contact center are tightly connected, and it breaks when they are not. Our integrated platform will let a supervisor see every interaction and let a human step in the moment an AI agent runs into trouble, which means the customer doesn't need to repeat their issue. Imagine a policyholder who contacts their insurer about a denied medical claim. An AI agent running on Five9 verifies her identity, retrieves the claims history, and confirms the denial reason within seconds. It recognizes a policy exception and routes her to the licensed claim specialist, selected by Five9 routing for skill and availability. During the wait, the AI summarizes the case, flags the relevant policy language, and surfaces prior contacts. The specialist picks up, already briefed, making the judgment call, and resolves the dispute on the first call.
Of course, our customers want to keep the cost of service down while also growing the value of their relationships. This insurance example shows how those goals can work together. AI takes on the preparation and the follow-up, which gives the specialist more time to explain the decision and help the policyholder understand her options. The insurer can handle the work more efficiently, and the customer gets personal attention when it matters the most. The efficiencies from this humantic capability allow customers to expand their reach with their base and allow for more interactions, which might not have been possible before. In keeping with this thesis, we saw strong expansion among existing customers this quarter, with AI contributing meaningfully to that activity. In particular, we're seeing healthy adoption of AI products we introduced last quarter, which started contributing to that expansion.
Taking a step back, the contact center industry is in early stages of a disruptive change that we believe will drive a significant investment cycle. Companies have historically managed customer service around cost and capacity. With AI making more high-quality interactions economically possible, we believe companies will begin investing heavily to earn their customers' trust, improve satisfaction and retention, and where appropriate, cross-sell or upsell them, driving new revenue streams. Each customer interaction is becoming more surface area for a positive experience. Nowhere is that more visible than in recent introduction of personal agents such as Muse, Dots, and Instinct. These agents remove some elements of customer support friction, and we expect they will drive incoming chat, email, voice interactions to the contact center. These services are early, and our customers are just beginning to ask how they should address them.
Although we are not counting on measurable financial contribution in the very near term, we firmly believe they are a sign of where customer interactions and the need to invest to deal with them are going. Over time, this is a large opportunity for us to monetize, both directly and indirectly. For agent-to-agent interactions, we're targeting a solution early next year. As personal agents make it easier to contact a business, we expect service requests to increase. Some will be resolved between AI agents. Others will require a person to handle an exception, make a decision, or resolve a dispute. Those escalations create another source of demand for human agents, even as AI handles more work routinely. That brings me to our second focus today. The next step is expanding the AI customers and use cases we can serve.
Shortly before this call, we announced that we are entering into a definitive agreement to acquire Level AI for a total purchase price of $325 million. We're very excited about this opportunity to bring a world-class AI-native team and their AI-first platform to augment our agentic offering and organization to better unlock AI TAMs that are transforming our industry. Let me walk you through some of those details. Level AI brings AI-native applications for Voice of the Customer insights, Agent Assist, and AQM, plus AI automation across voice and digital channels. Level AI's proven AI-native platform and applications will expand the ways we can win new customers while also accelerating our time to market. We've worked closely with Level AI since 2022, so we know the products and the team well.
Its applications are already integrated with our CCaaS platform, giving us direct experience with the technology in customer environments and a strong foundation to build on. Combining Level AI's AI-native team with our own AI talent will help us accelerate development and deliver a more complete offering sooner. Another important reason for this acquisition is the opportunity to bring standalone AI capabilities to enterprises using other contact center platforms, both in the cloud and on-prem. That includes enterprises that want to use AI to improve customer service but aren't ready to move their contact centers to the cloud. Level AI lets us serve those customers without requiring them to first buy our CCaaS platform. Historically, we've led with CCaaS and brought AI into that relationship, and we'll continue to lead with CCaaS for customers ready to migrate.
Level AI adds another starting point, addressing a customer's immediate AI needs while its existing contact center stays in place. That expands the customers we serve today. Standalone AI is an attractive business in its own right, whether or not those customers eventually adopt Five9 CCaaS. It also gives us a relationship to build on. When an on-prem customer is ready for the cloud, we can compete for that business, having already demonstrated the value of our AI. Five9 brings the sales force and customer relationships to take these products to a much larger market. We can introduce them to our existing customers and sell them independently to enterprises using other contact center platforms.
For customers of Five9, our goal is to meet their humantic needs by connecting these capabilities more closely with our voice routing and orchestration platform, so teams can use conversation insights to guide coaching, identify opportunities for automation, and improve workflows across human and AI agents. Level AI reported standalone annual recurring revenue of approximately $20 million as of October 31st, and grew approximately 50% year-over-year. For planning purposes, we're anticipating they'll grow ARR at a similar level again over the next 12 months, excluding any upside from synergies. Taken together, we believe the agreed purchase price represents an attractive value for Level AI's standalone business, with additional upside from bringing its products to market through our sales force. As we grow that business, we expect the acquisition economics to become even more attractive.
That's how we approach capital allocation, discipline on the price we pay, and a clear plan to create additional value for shareholders. We expect Level AI's strong growth to continue, and together, our CCaaS business and expanded AI portfolio position us to serve customers regardless of how contact center technology evolves. Importantly, we expect the addition of Level AI to accelerate our agentic product suite in totality. This combination strengthens our confidence in our ability to deliver double-digit revenue growth over the medium term. In closing, we're building Five9 for a larger volume and wider range of customer interactions, strengthening the connection between people and AI, and expanding customers we can reach. Our focus is on helping enterprises keep the cost of service down while growing the value of their customer relationships and delivering a clear ROI on their spend with Five9.
With that, I will turn the call over to Bryan to review our quarter, our outlook, and the financial overview of the transaction.
Thank you, Amit. Good afternoon, everyone. I will walk you through our Q3 financial results and then provide guidance for the remainder of the year. Q3 revenue was $323 million, up 13% year-over-year, and above the high end of our guidance. Subscription, telecom, and professional services represented approximately 83%, 11%, and 6% of total revenue, respectively. Subscription revenue grew 16% year-over-year, driven by CCaaS growing 8% and AI revenue accelerating to 91% year-over-year growth. AI revenue was approximately $45 million, or 17% of subscription revenue, up from 10% a year ago. Several customers deployed earlier and at a larger scale. The earlier deployments benefited the timing of revenue, while larger deployments increased their scale. While we acknowledge that these strong ramps will create difficult year-over-year growth comparisons for 2027, we do feel good about the near-term trajectory.
Consequently, we now expect full-year 2026 AI revenue growth of at least 75%, up from our prior outlook of at least 60%. Concurrency count grew relatively in line with CCaaS revenue growth, and we expect that trend to continue in Q4. LTM subscription dollar-based retention rate was 109% in Q3, up from 107% in the prior quarter, driven by both increased seat adds and AI deployments from the backlog. We anticipate this key metric to remain flat to one point up in Q4, driven by the timing of our backlog converting to revenue. Now turning to profitability. Adjusted gross margin in Q3 was 62%, compared to 63% in Q3 2025. The year-over-year decline reflects timing-related costs with one of our largest partners, and we anticipate this to start reversing in Q4. In the fourth quarter, we expect adjusted gross margin to continue moving higher.
Adjusted EBITDA margin in Q3 was 25%, compared to 25% in Q3 2025. Non-GAAP EPS was $0.82 per diluted share, compared to $0.78 per diluted share in Q3 2025. In terms of cash flow, cash from operations was $58 million, or 18% of revenue, and free cash flow was $43 million, or 13% of revenue. We ended the quarter with approximately $698 million in cash equivalents, and short-term investments. We remain on track for purchase of PP&E to come in at approximately 3.1% of revenue for 2026. As a reminder, this is temporarily elevated by approximately one percentage point above last year's rate to support our global data center refresh in 2026. We completed a $90 million accelerated share repurchase on September 30th, buying back approximately 3.5 million shares at an average share price of $25.97.
We have $200 million remaining under our authorization, and we plan to deploy it opportunistically. Before turning to guidance, some comments on Level AI. The acquisition is for $325 million in cash from our balance sheet, subject to certain purchase price adjustments. We expect this transaction to close during the fourth quarter. Our Q4 and full year 2026 guidance exclude Level AI's revenue and operating contribution. When we report our fourth quarter results, we will discuss the financial impact of the Level AI acquisition. As Amit mentioned, Level AI's standalone annual recurring revenue of approximately $20 million grew approximately 50% year-over-year, and we expect it to grow at a similar level again over the next 12 months, excluding any upside from synergies, representing a transaction multiple of approximately 11x based on next 12-month ARR.
For bottom line, it is still early, and several items could change, but our preliminary work indicates approximately one point of dilution to our adjusted EBITDA margin in 2027. However, we plan to bring Level AI's EBITDA dollars to break even in approximately six months and EBITDA margin to accretion in 2028. We will share more details when we provide formal 2027 guidance in conjunction with our 2026 full year results in February. We are moving our investor day to 2027 so that we can present our strategy and long-term financial outlook for the combined company. We will confirm the revised timing when available. Moving to our guidance, which as I mentioned, excludes any impact from the acquisition. For total revenue in the fourth quarter, we are guiding to a midpoint of $331 million, with a range of $328 million-$334 million.
Within that, we expect a sequential step-up in subscription revenue, flat telecom revenue, and a sequential decline in professional services. For the full year 2026, we are guiding total revenue to a midpoint of $1.272 billion, with a range of $1.269 billion-$1.275 billion. The midpoint is $6 million above our prior guidance, reflecting our Q3 outperformance and an increase in our implied Q4 outlook. The conversion of backlog to revenue continues to support revenue guidance in the fourth quarter. Further, seasonality in the fourth quarter is expected to be muted, similar to last year. For non-GAAP EPS in the fourth quarter, we are guiding to a midpoint of $0.90 per diluted share, with a range of $0.88-$0.92 per diluted share.
Our guidance for 2026 non-GAAP EPS is $3.19 per diluted share at the midpoint, with a range of $3.17-$3.21 per diluted share. Note that the midpoint is $0.07 below our prior annual guidance of $3.26, primarily due to timing revision and planned org structure changes, and a higher diluted share count resulting from a higher-than-expected share price. We have included a bridge in the appendix of our investor presentation that quantifies each driver. For the full year 2026, we continue to expect our adjusted EBITDA margin to be at least 24% and annual cash flow to be approximately $175 million. I would like to note that the revised timing of planned org structure changes also affects Q4 adjusted EBITDA margin, although we still expect a sequential expansion.
Please refer to the presentation posted on our investor relations website for additional estimates, including share count and taxes, as well as GAAP to non-GAAP reconciliations. With that, I'd like to open the call for Q&A. Operator, please go ahead.
We will now begin the question-and-answer session. If you would like to ask a question and have joined via a panelist link, please use the raise hand icon, which can be found in the black bar at the bottom of the webinar application screen. If you have joined via telephone, please press star nine on your keypad to raise your hand. When you hear your name called, you will be prompted to unmute your line and ask your question. If you have joined via phone, please dial star six to unmute your line. We will now take a moment to allow the queue to form. Our first question comes from Siti Panigrahi with Mizuho. Siti, please dial star six to unmute your line and ask your question.
Oh, can you hear me?
Yes.
Hello? Okay, that's great. Thank you. Thanks for the question, and congratulations. Good quarter. Look, Amit, you entered the second half with essentially no dependency on incremental new logo to go get, mainly to achieve your 2026 guidance. You now have several large enterprise ramps, continue AI momentum. As we look beyond the strong backlog driven, how much visibility do you have today into sustaining this kind of double-digit revenue growth? What are the biggest variable between roughly 10% growth versus something closer to mid-teens, mainly as you're looking into 2027?
Hey, Siti. This is Bryan, and I am happy to answer that question. Today, we are not going to give formal guidance for 2027, especially given the acquisition that we just announced. But I can give you some high-level commentary to help you with that. There is obviously a lot of momentum in the business, and what we are thinking right now is that consensus, if you look at it, is on the revenue for organic basis is at $1.39 billion, and we believe that is a reasonable starting point for next year. To kind of look at the shape of the curve, as we always say, Q1 is relatively flat sequentially from the midpoint of Q4 guidance we have provided. Then it will step up slightly in Q2 and more so in the second half.
From a bottom line perspective, again, on an organic basis, we are guiding to at least 24% EBITDA margin this year in 2026. We will continue to focus on gradually expanding that throughout 2027. But one thing I do want to point out is that in Q1, the expenses are always significantly higher because of all the seasonal items that kick in, like the FICA reset.
Okay. That is helpful. If I could ask a follow-up. Amit, Level AI acquisition. Why Level AI, and does this replace your current IVA and QM roadmap that you have, or sit alongside it? What kind of overlap you have in terms of customers that Level AI has?
Yeah. Thank you for the question, Siti. Look, let me just start by framing a little bit, and then I will tell you how Level AI fits in. But, if you recall, I have always been talking about the CCaaS market is about $25 billion in size. The CX AI market is expected to be about $25 billion by 2030. A few years ago, we started building AI and said, Hey, this market is now, we are going to attack both parts of it because we believe in the humantic nature of customers, building AI with CCaaS and kind of moving towards a platform. A lot of what we did, Siti, when we first started, was to build AI that actually worked with our platform. And you have seen the success of that.
The numbers we are putting up on the AI growth, a lot of it is coming from the base. Our angle of attack was always CCaaS first and then AI. What has since happened is we have seen a lot of prospects coming to us, especially as we have said, Look, a big chunk of the market is still 40%, 50% is still on-prem, going to make the move to the cloud. A lot of them have come to us and said, look, I would love to go test a full stack of AI on day one. What Level AI gives us now is a full suite of standalone AI products with day one reference ability. They have a lot of large customers already working with us. They were available through our app store, integrated with our CCaaS, so we know what it would do.
Now this reverses it for us, where it gives us a second line of attack, which is AI first and then CCaaS second. I think no one better to talk about a humantic conversation with a customer than one that is actually offering a full stack AI plus the CCaaS and being able to talk about the interchangeability and the one plus one equals three there. The second part was, they bring a robust team to augment all the stuff that we are already doing and accelerate our roadmap by about 12 months. When I add all of that together, for me, the ROI on this pencils out from day one.
I think when you look at this two, three years from now, I think we will probably look back and say we bought this for a steal, just given what I think it can do for the business. In terms of the overlap, yes, there is a little bit, but that is the work that is going to be done in the next, before close, to figure out, like, how we want to attack this. I will just say end-to-end, it gives us more capabilities, and it gives us a new line of attack into the market, which I am really excited about.
Amit, that is a great color and ties well with your 91% AI growth. Congratulations.
Thank you, Siti.
Our next question comes from Arjun Bhatia with William Blair. Arjun, please unmute your line and ask your question.
Thank you so much. Amit, maybe we can talk about the AI growth acceleration. It seems like it was obviously very strong this quarter. I'm curious, just as you think about that, what's changing? Does it seem like we're approaching a tipping point in terms of adoption? Because the numbers are clearly kind of pointing to something there. When you think about the broader customer base and we start to look into 2027, are there still roadblocks that you see customers trying to battle in terms of AI adoption going into next year and beyond?
Well, thank you for that question. Look, I'll just start by saying, and I'm sure Bryan will add some more color on the customer size and what's happening there. But, look, in general, we've been saying this, customers have moved past the POCs and now are in adoption. I think what you are seeing from us, a lot of this growth that you're seeing is coming from the base. It's also coming from customers who've adopted one product and two and keep expanding and building new use cases once it's been proven out. That is driving a big piece of it. Look, in terms of inflection, I mean, we're already at now 17% of our subscription revenue is AI. It's up from 10% last year. This is now becoming meaningful. I don't think this is a sideshow for anybody.
This is really the way contact centers and service centers and others are starting to deploy. What is also happening is our messaging of humantic is resonating. Because in this case, we also see a lot of customers. When I came to this company, there was this whole fear that all humans are going away, it's going to be replaced by AI. We were always the number one to say, "Interactions are going to go up. The economics of the contact center are going to get better with AI, and it's going to drive more interactions." We're seeing that. A lot of our customers in regulated industries are maintaining, in some cases adding heads, but also deploying AI more fully across their stack. I'll just close with this.
The gap for us was always on the prospect side of things, where someone was saying, hey, I am not ready as yet to make a switch to the cloud. Look, it's 2026, if you're still on-prem, you really want to be on-prem. But there, for us, those customers have come to us and said, give us a full-stack AI solution. Let us go test it out. Let us see the full value. Then, it'll follow the cloud conversation of CCaaS comes second. This is where Level AI fits in for us, and really excited about what that opportunity brings. Brian, I'll turn it over to you.
Yeah. Arjun, the only thing I'll add here is that, we've been talking about the strength of our backlog converting to revenue, and it's really on both the CCaaS and AI side. To Amit's point about the importance of Humantic, the step-up as anticipated, CCaaS went from 7% growth to 8% in Q3, and AI went from 78% to 91%. It's really a reflection of the business that we've won that we're now starting to convert into revenue.
That's very helpful. Thank you, guys.
Our next question comes from Raimo Lenschow with Barclays. Raimo, please unmute your line and ask your question.
Thank you. Congrats from me as well, and great acquisition. Two quick questions from me. We obviously have some very large customers at the moment that are going live or in the process of going live. Can you talk a little bit about the progress there, and also what you're seeing in terms of early pipeline from other guys that have looked at what you're capable of doing now? Level AI is really good, but you also partner with other players in the market. Does that kind of create some headwinds on some of those transaction or deals that you were working on? Just to kind of double-check and that we get all the numbers right. Thank you, and congrats again.
Thank you, Remo. Look, I'll answer the second part of the question, then I'll turn it over to Brian. Look, one thing for us, and we've always said this, is we are and will remain an open ecosystem, because our belief is, in this space, it's really hard to come in and dictate the full stack, and especially when you're working on an interaction layer, which I kind of now, with all the AI progress, I don't view ourselves as a standalone CCaaS company. I view ourselves as an interaction layer company. You're going to have to work with different pieces to complete that full stack. So, that's one. Two, there are some use cases, Remo, that we see, which is vertical-specific or very niche-specific.
The other thing that you've heard me say is, look, we are going to be very diligent with our dollars and where we spend, and I am not going to go build everything for everybody. So, for that viewpoint, partnering makes sense, and serving up those sorts of solutions makes sense. With Level AI, for my view, what it does is it completes a big part of where I think the opportunity is. It gives us the opportunity to present a faster, more robust roadmap to our existing customers, as well as prospects that want a full stack. So, that should just add more firepower to us, and there will always be gaps around that we will fill with partnerships.
Yeah, and I'll just add on to your first question, Raimo. There are a number of large customers, but also mid-sized customers, and across new logos and install base that are in our backlog that's converting to revenue. The ramp's going right on track for vast majority of those. There's puts and takes naturally. Then, in terms of the pipeline that you talked about, especially around our partnerships with someone like Google, for instance, the pipeline remains robust. There are a lot of large enterprises, opportunities looking ahead, so we remain optimistic about kind of that long-term growth opportunity ahead of us.
Lovely. Thank you. Well done.
Our next question comes from Elizabeth Porter with Morgan Stanley. Elizabeth, please unmute your line and ask your question.
Great. Thank you so much for the question. I wanted to circle back on the idea of customer agents. Appreciate you guys aren't assuming any financial impact in the near term, but just as customer-facing AI Agents like Muse and Instinct increasingly act with businesses on behalf of customers, what are you seeing or what's the thought process on changing contact center volumes and the overall opportunity for Five9, and how does Five9 really capture the incremental revenue as more interactions are shifting to AI over time?
Elizabeth, I'll take that one. Thank you for the question. Really good one, because it's on top of mind for us. Look, I will say, when I joined this company, we were always very focused on, hey, interactions are going up. AI is going to drive more interactions for this business, not less. We saw the first phase of that through better economics in the contact center, reaching more customers with the services provided. Personal agents in this new wave is phenomenal because it has just expanded this interaction layer significantly. You are right, it is early days. That is really exciting for us. In fact, I think it makes platforms like ours more relevant, not less, because of all the capabilities we bring. Look, there are two ways we will monetize this opportunity. One is direct.
We are working on a product, and we can spend a little time talking about that, but eventually, we said we're bringing out our agent-to-agent and routing and all those capabilities for personal agents in the early part. That's frame one that we'll actually navigate and monetize. Then the second one is breakage. Look, when you have millions and millions and millions of interactions, you're going to see breakage that happens, and that will lead to additional phone calls, additional chats from the human that originated it, and that's obviously something that we monetize today. That is where we feel the world is going. Why I think a platform is really important here is because what we do is we cobble together voice, AI, digital orchestration.
An agent comes in, the customer has to be able to decide how you want to route it, how do you authenticate it, how do you check for permissions, and then what channel do you want to use. Effectively, you're going to go down the same process with an agent that you would when a customer receives an inbound call or a chat and decides how you want to handle it. So platforms like ours, I think, just have a more favorable position in that environment.
Great. Then just as a follow-up, maybe one for Bryan. You noted more muted Q4 seasonality. Can you just unpack a little bit of what you're seeing today that would drive that more mutedness going into next quarter? Thanks.
Yeah, absolutely, Elizabeth. We actually did a survey back in July with our top seasonal customers, as we do each year, and they cited muted seasonality just like last year. It is really dependent on the uncertainty of the macro, the demand level that they are seeing, and that has not changed. Going into Q4, what we are guiding to is another muted season, just like it was last year. What is really reflected in the guidance is the timing of the backlog converting to revenue, because every customer in that backlog has a unique schedule ramp, and that is what is reflected in our Q4 revenue guide.
Great. Thank you.
Our next question comes from Scott Berg with Needham. Scott, please unmute your line and ask your question.
Hi, everyone. Nice quarter, and thanks for taking my questions. Amit, your announced acquisition of Level AI was interesting to me because it is the second native AI company to be acquired by a CCaaS vendor this week. Timing is everything, I guess. What is the acquisition or the consolidation of a vendor like this, I guess, what does it do in the market here today? Is this a reflection of customers are struggling to integrate or to leverage these native platforms on their own, and they are really best fit in this full-stack scenario that you guys are all trying to help put together? Just help us understand, is this a new trend, and is this the, I guess, the demand path that we should be seeing from customers going forward?
Yeah, look, I won't speak on behalf of what other companies are doing because I'm not clear all the way on their thesis, but I can tell you ours. For us, we've been saying this for a while. We've been talking about the humantic ecosystem. For us, it is really about how do you build, put together, and drive very clear back and forth for AI and humans, digital, voice, AI orchestration. Those are the four components. And I think we've cited some examples of what happens when it actually all sits on the same platform where routing is controlled by us, where telephony is controlled by us, and the advantages that gives you on latency and the ability to actually manage the interaction, not lose context as information is going back and forth.
So for us, building out AI, putting it on a platform is part of that exercise and the direction we're going with our humantic mission. Level AI, it just gives us, obviously beefs up our roadmap, but more importantly, allows us, as I mentioned, just another line of attack into the market, which is, hey, you can actually come and use a full-stack AI solution, and when you are ready, who better to talk to you than a company that actually does CCaaS and AI to show you what the power of 1+1=3 is, versus you assuming that. That's how that fits into us and where I think this market is going from our lens.
Yeah, that's very helpful. Then, Bryan, the timing revision that you talked about and some of the org change stuff, I guess we typically don't see this very often. Usually, these things may be accelerated as much as anything. Help us give this maybe some background and why this is getting pushed out, and will this all be completed in Q1 next year, or will there be, I guess, further delays going into 2027? Thanks.
Yeah, Scott, absolutely. We're going through a lot of different initiatives in terms of cost savings, org design changes, and this was one that we originally planned for Q4, but it's purely a timing matter. We're going to continue to focus on executing on it in the earlier part of 2027. But the most important part here is that for 2026, we're still guiding to at least 24% EBITDA margin, which is what we've been guiding to throughout the year. Free cash flow, we're still expecting approximately $175 million for 2026. And for Q4 EBITDA margin, we're continuing to expect that to move higher in the fourth quarter as well.
Helpful. Nice quarter. Thanks again.
Our next question comes from Catharine Trebnick with Rosenblatt. Catherine, please unmute your line and ask your question.
Yeah, thank you very much for taking the question. Just a couple of ideas on this acquisition. Are they going to stay as a separate sales force or going to integrate the sales force? How are you going to do the quota for them? It will be AI-based or it will be CCaaS-based? Thanks.
Hi, Catharine. Thank you for the question. Look, all good questions and things that we are in the works to get aligned. As we have more on that, I will share with you. For us, this is really about accelerating our AI capabilities into the market. Today, our sales team already carries quotas for CCaaS and AI. For me, it is really about how do they help us get better in terms of driving more AI into the acquisition base. More to be sorted out there, and we'll keep you posted, but I'll leave it at that.
Okay. Just a quick follow-up. Is some of the customers coming from non-Five9 or on-prem contact center platforms like Genesys or Avaya? Thank you.
A very small portion of Five9's customer, of their revenue is with Five9. A lot of it is additional customers. And they do have some sizable on-prem customers that are using them. For example, one is a Fortune 100 insurer that is on an on-prem solution using Level AI's offering to augment and help all their humans in their centers. For us, that was the referenceability. We know it works there at scale in enterprise settings, and we get a lot more phone calls for those sorts of questions. I think it just works well.
All right. Congratulations. Thank you.
Thank you, Catharine.
Our next question comes from Samad Samana with Jefferies. Samad, please unmute your line and ask your question.
Hello. This is Joe Picora on for Samad Samana. Thank you for taking our questions. First, I wanted to revisit the capital allocation priorities after completing the accelerated buyback and, of course, the Level AI acquisition upcoming. How are you thinking about future capital allocation priorities between buybacks and M&A from here?
Yeah, Joe. It's a good question. We have a very robust framework for evaluating all of our capital allocation alternatives. As you said, in the last 12 months, we bought back about $150 million. Given the long-term growth opportunity and the disciplined price that we paid, we believe the acquisition that we made at Level AI was the optimal use of cash at this time in terms of creating additional value for shareholders. Having said that, we are generating strong cash flow, free cash flow. I said $175 million for 2026, and we expect that to continue to march upward. We have a $200 million share buyback program in place. We'll continue to be opportunistic in terms of executing against that.
Understood. Thank you very much. Second, I wanted to again revisit the Level AI acquisition. Appreciate how it can be another point of attack. At a higher level, are there certain sectors or verticals that Level AI has been able to better penetrate that can help support the pipeline going forward that Five9 may have not had as much success in?
The answer to that is, what I actually looked for is, we've already declared that we are going to go after regulated industries. We think that's where the complexity is, and a big part of our revenue base sits there. When we looked at Level AI, the thing for me that was important is, do they have all the prerequisites to go and quickly let us go do that? Now that we've worked with them and partnered with them for a while, they have all the regulations, GDPR, HIPAA, PCI, et cetera, et cetera. It really lets us stay focused on our plan and work to get better outcomes from it. That was really what was important.
Look, the other piece I will just say on Level AI is, I've been here in seat eight months, and we didn't go out and rush to go pick up the first AI company that existed. We've been very thoughtful, waited for the right one that actually fit our strategy at the right price, and it checked a lot of boxes. I've known the founder, Ashish, for a while, and we have a similar vision of a humantic universe. Just wanted to add that in, that this has been a very thoughtful kind of move at the time we've made it.
Awesome. Appreciate it, guys. Thanks again.
Our next question comes from Terry Tillman with Truist. Terry, please unmute your line and ask your question.
Great. Good evening, team. Connor Passarella on for Terry. Appreciate you taking the question. As AI deployments scale, especially with some of the larger customers, could you maybe just talk a bit about the evolution of the services investments needed to get customers live and what you're seeing as the delivery model is becoming more repeatable and less resource-intensive? Thank you.
Yeah. So, definitely. First of all, I'll talk about the AI portfolio. The three biggest products that we have there that are contributing to that 91% year-over-year growth are AI Agents, and then Agent Assist, and AI Insights. If you look at the traction for this quarter, for instance, AI Agent was the biggest ramp that drove the acceleration. We are, from a resource perspective, always have the right level to go as fast as the customer wants, and also, we're constantly working on improving our technology to make that implementation faster and faster. So, it's a long process to get there, but right now, we're already showing good signs. As you can see, a lot of these deployments for the last three quarters, when we talk about AI ramps, they're coming in earlier than anticipated.
That's partly also the efficiency that we're getting on the deployment side, as well as the customer being ready to go faster. So, we're feeling good about that side of the business.
Very helpful. Thank you.
Our next question comes from Jackson Ader at KeyBanc. Jackson, please unmute your line and ask your question.
Great. Hey, guys. Thanks for taking our question. On the acquisition, I'm just curious whether I mean, I'm sure that you've talked about a lot of product roadmap, both on your side and on the target side. I'm curious where there is going to be maybe some TAM expansive R&D opportunities, and then maybe other opportunities where it's like, you know what? We're working on the same stuff. We can kind of consolidate those.
Yeah. So, look, thank you for that question, Jackson. On the TAM expansion opportunities, Level AI brings four to five very key products that are already built out. For us, the TAM expansion is it just gives us more firepower with those products that either were super nascent with us or not even built, right? That is one piece of it. I think that's really the main piece. Then to the question around, does it allow us to retire some part of our asset, et cetera? Those are things that we have to go through. But at the end of it all, when I look at what they bring and the feature functionality and the additional product fully built out, the roadmap is accelerated. Net-net, that's the real impact.
I think the only thing I'll add there, Jackson, is that if you look at their ARR as of October 31, we mentioned it's $20 million, and it's growing very quickly at 50% on a standalone basis, and we expect that level to kind of continue next year, excluding any synergies on that front. But one thing I did want to clarify is that we are talking about ARR, so of course, naturally when you're modeling, there's going to be a lag to revenue there.
Sure. That makes sense. Just real quick on backlog remaining performance obligation. I know stuff can swing around. It depends on a bunch of different things. RPO up a bunch again in the quarter. I am just curious, as we kind of head into the fourth quarter and into 2027, anything you can share with us on kind of like pipeline or coverage ratio relative to fourth quarters in the past? Thank you.
Got it. So, Jackson, in terms of RPO, as I have always said, it is not a good indicator of future revenue because any customer contract that is 12 months or less are excluded. Having said that, you are right. RPO is now at $1.6 billion. It went up 11% quarter-over-quarter, and that was primarily driven by one of our largest customers who actually committed to a much longer term with us, and therefore it was now counted in the RPO. It is not a tool that we use internally for forecasting, so I just wanted to emphasize that point. From a pipeline perspective, we feel really good about it. It stays very strong. A lot of the partner momentum that I talked about with Google, that is a real solid, robust pipeline for us. So, we are feeling good on that front.
Got it. Thank you both.
As a reminder, if you would like to ask a question and you have joined via the panelist link, please use the raise hand icon, which can be found in the black bar at the bottom of the webinar application screen. I will pause for a moment to ensure there are no further questions. This concludes our Q&A session. I will now turn the call back to Amit Mathradas for closing remarks.
Thank you all for joining us today. The leadership team is working well together. Our AI offerings are performing, and the way customers think about their contact center lines up with our Humantic CX vision. We believe this is not an AI or human choice. The next horizon needs a platform that connects AI and people across every interaction. Customers asked us for a standalone AI and demanded that we build and offer a solution, and Level AI is that answer. The market is moving fast and represents a multi-billion dollar opportunity, so we will stay focused on the opportunities in front of us. Thank you for joining us today.
The call has concluded. Thank you for joining. You may now disconnect.