Five9, Inc. (FIVN) Earnings Call Transcript & Summary

Welcome to the NASDAQ Investors Conference in partnership with Jefferies. Let's now go live to London.

Before we kick off, Barry, I know you wanted to give a Safe Harbor agreement because you're going to give forward-looking statements and mine often are wrong. So I know you're going to give yours just in case as well. I guess things are different.

So hello, everybody. Yes, before we begin, today, we'll be making forward-looking statements about events and trends that can affect the industry, the company and its operations. We will also make forward-looking statements about product development, about AI and automation and about growth prospects. These statements should not be unduly relied upon by yourselves, and we undertake no obligation to update them. For factors that could cause such a difference -- material difference between what we -- forward-looking statements and actual results, please refer to our 10-K and 10-Q filed with the SEC under the caption "Risk Factors".

Great. So with that, AI is all of the rage right now in software. And we're lucky to have Jonathan with us, who is the CTO and he's been a long-time industry veteran in the space. And he's going to present a couple of slides of just where Five9 fits into the world and so, Jonathan, I'm going to let you kick it off.

All right. Sounds good. All right. So, first question is, who's Five9 and what do you do? I'm going to answer that question. We're a software company. We sell software to enterprises. It's a cloud software company, so we're Software-as-a-Service. What does our software do? It helps customers manage their inbound and outbound communications with their own customers, contact center, call centers. So being on calls, chats, e-mails, we sell the software platform to enterprises. Traditionally and typically, it replaces an existing old fashion on-premise thing that they bought from vendors who are now largely either out of business or they have end-of-life those products and these customers need to move to cloud platform. We're one of the leading cloud platforms for that capability. So that's what it is to sort of go just one click deeper on this. The way to think about it is we're a platform in the middle. And customer interactions, we call them, come to our platform. This is a phone call that you make. So if you pick up your mobile phone, you dial 100 whatever, it comes into our system. If you click on a web chat, which is on the web page, comes to our system if you send an SMS, an e-mail, WhatsApp. So our system accumulates all these interaction types across all these different channels, and it comes into our platform, we process interaction. So that's the first part as we do these channels and interactions. The second part is our platform processes them. What does it mean by processes them? It holds them so we can do things like route it to the right place, we can report on it, we can connect it to different agents or voice bots or chatbots as the need arises, we can automate parts of that interaction, which I'll talk about a little bit more. So we have a whole bunch of capability. We're going to go a little bit deeper in that in just a moment. And then the third part of it is that we integrate with third-party systems. So contact center is all about integration. So if you call up and ask like what's the status of my package, the agent needs to be able to answer that information to connect to the packaging system. If you want to build a voice bot or a chatbot that knows about the status of the package, we have to integrate with them. So we have a platform that provides these integrations using APIs into these enterprise application systems. And the last part was engines. Five9 has been an AI company for a while I'm CTO and Head of AI. I've had that title for my entire career 4.5 years here at Five9. And we sell software that automates those interactions also. So if you call up or chat, you get answers by voice bot or chatbot. Our system does handle that interaction. It's part of what we sell. But we don't build the raw natural language understanding or speech recognizer capabilities. We plug into third-party engines like Google and now Open AI for those technologies. So just one click deeper, there's a lot of things on this, and you probably can't see it either, but that's okay. So let me sort of take you through a typical flow of what happens to our system. So let's say, you are a consumer and you call up. So you pick up the phone and you dial a phone number and that routes into our platform. And what happens is it hits the first component, which is our routing engine. It's a core part of what we do, and it makes a decision where should this call go? Should it go to you because you're the right agent for this customer? Should it go to you a different agent because you're not -- you just got off the phone and you haven't handled many calls yet? Should it go to a voice bot because this is like a customer that does spend a lot of money with us and we want to just send them to the voice bot? This logic that the customer can configure is part of what a routing engine does. In this example, what's going to happen here -- is it gets routed to a virtual agent in this case. So we sell those. We acquired a company 2.5 years ago called Inference, which is the market leader in virtual agents, voice bots, chatbots, that is able to answer and integrate with systems to answer the inquiry. And to do that, it leverages a lot of these third-party engines. So speech recognition, to understand what the user is saying, speech generation to speak responses, natural language understanding to process the meaning of what the customer is saying and in order to take further action with the thing. So all these green components that's what our platform is good at plugging into and swapping out these engines based on the best cost performance rate. Then the next thing we do is we want to do something. In this case, you have to identify the customer, a customer put in their account number, we got to look them up in the case system. Our platform allows for integrations with third-party systems. We have a great like drag-and-drop builder that allows us or our customers to build these integrations, so we plug into the CRM, maybe we plug into the marketing system, and we learn that we just sent you an e-mail recently. That's why you called back. So the IVA can say, oh, thank you, Bob, for calling. Are you calling in response to the e-mail we sent about the new promotion for the product. So we plug in the marketing system for that. . All right. Great. Now in this particular case, you've got some complicated issue and you want to transfer it to a live agent. No problem, our platform can do that too. So we transfer the call to a live agent. So our system has a web app that sits on the agent's desktop. That's what the agent interacts with that allows him to do all the standard call stuff like call transfer part but also shows some interaction about you, the customer, who's calling up the contact center, back-end systems, see the e-mails, chats, the history of the conversation, guidance on which you should do next are all provided in the UI that we deliver to the live agent. We also do things like assisting the agent, providing them next best actions or AI systems plug into third-party engines like LLMs are able to understand the call, give the agent suggestions. And then at the end of the call, we can summarize the call for the agent, so that agent can review it, approve it and upload it into the CRM system. Then we might have workflows. So example, let's say, you were talking to the agent and you expressed interest in a product the agent says, oh, can I send you an e-mail about that? You say, sure. Our system can plug into third-party systems for sending an e-mail or an SMS for example after the conversations, don't ask you fill out a survey or ask if you have follow-up questions, so you can continue the dialogue. So that's all more channels that plug into our system. And at the end of the thing, it's all about data. Contact enter is a heavily data-driven part of the enterprise software market. So we provide graphs, dashboards, visualizations that lets our customers answer questions, like what percentage of my calls are you able to be offloaded to bot? How efficient are my agents? How many agents do I need? All these kind of questions. How many calls was I handling? How many calls versus e-mails do I have? This is what we provide analytics for. That's it. That's basically what our platform does. I think that was under 7 minutes.

Exactly well done.

All right. Thank you.

Man, I wish I could be that succinct and timely. So thank you, Jonathan, for that. I really appreciate that. Maybe I'll kick off with a question for either you or Mike, maybe how do you see ChatGPT, which has been heavily discussed and LLMs evolving over time as customer service and customer engagement seems to be the area that's going to be the most affected by the rapid evolution of technology.

Maybe I'll take that one. So first thing I'll say is large language models, ChatGPT, OpenAI, these things for us are accelerants for our business. Five9 is in the business of selling software that helps customers deal with their inbound request for information and interactions with customers, the more we can automate of those, the more money we make. We see large language models ChatGPT is doing a way better job at that than the prior generation of technology. So it's really good for us. Now we're in the very early stages of this tech. It's like 6 months -- I mean it's been in the news for 6 months. It's been around maybe a year. The first paper was actually published 2 years ago as it takes a while. So what we think is going to happen in the next generation is actually not so much about just the evolution of the large language models, which will get more accurate, less hallucination, more capable, lower cost, it's about building applications on top of it. That's how things are going to evolve. So if a brand and a customer, an enterprise wants to use one of these large language models to handle customer interactions, they come to a vendor like us, and we'll integrate that large language model, and it becomes customized for that customer -- for that enterprise using something called prompt engineering. This is the new hotness. You're all going to hear about prompt engineering in and out over the many years. And to do that, our platform takes context about the customer, knowledge -- special knowledge about the company and their practices, and it feeds that into the LLM via the prompt. And so our software is in the job of prompt engineering. That's where we think we're going to see a lot of innovation and expansion in the coming years.

So that's a great way of thinking about what's under the hood. Mike, let's zoom out, it's been your baby for a long time. What products this Five9 have today on the AI side? And how should we think about maybe where AI sits inside of the product portfolio that customers are actually asking about?

Yes. So customers, first and foremost, Samad, are asking about how do I replace these legacy on-premise solutions like Avaya, Genesys and Cisco that they've been basically operating their contact centers on for years and years. And they know that a lot of these solutions are being end of life. They've got vendors that are in and out of bankruptcy. They've got to get off those platforms. But they also -- their first and foremost priority is customer experience. That's what we do for a living our software, helps our enterprise customers, reimagine their customer experience for their end consumer, if you will. And AI and automation is a huge part of that now. . It's always been part of our strategy. In fact, we've acquired Inference 2.5 years ago to basically give us that entree into AI and automation. What they're now realizing is not only can I move off of these legacy systems, improve my customer experience by moving to the cloud, but I can also take advantage of this AI and automation opportunity, which has very tangible ROI. And Jonathan already explained kind of part of how that works, but it's an incredibly powerful catalyst for the shift to the cloud.

So that's -- I have a follow-up to that. AI feels like this tectonic shift, right, where you guys have been working on it for years. And suddenly, in the public, there's a shift of plates and now there's an earthquake, we're all talking about it, right? And I bring that up because you mentioned leaving legacy on-premise systems. Can you maybe just help us understand to take advantage of something as complex and robust as AI. Can someone stay on-premise to do it? Or is this what makes you go to the cloud for the holdout?

It's extremely difficult. In fact, you pretty much, from a practical standpoint, have to move to the cloud to take advantage of this. And we've made the analogy many times, but a good way to think of our platform versus LLM is really the analogy of a jet airplane. We are the airplane, the LLM are the jet engine. And so what these engines, all the innovation that's occurred all this revolution that's happening recently is really just empowering our aircraft to fly faster and fly further by utilizing these better engines. And those engines are going to continue to get better. It just essentially allows our end customers to deliver on our platform, all the applications necessary to deliver that great customer experience.

Understood. And I don't know if this next one is for Barry or for you. But since Barry, you're the numbers guy, I'll leave it for either of you. What percentage maybe of the RFPs that you're having with customers now are including AI solutions? Or just how are -- I guess, maybe customers asking about it when you're thinking about new deals and new contracts?

Yes. Thanks Samad. We track that very closely. In the fourth quarter of last year, 41%; in the first quarter of this year, 42% of the RFPs that we received had AI and automation is either the main driver prompting that or an auxiliary driver. So pretty high proportion and gently increasing.

And then maybe a follow-up, just historically, the company is priced on a seat-based model. There's a lot of debate that you could see maybe the number of agents change if AI replaces humans. I'm not sure I'm there. I've never called a contact center and been put through because of unusually low call volume. It's usually the other way around, right? When that happens to anybody in the audience, please let us know. But how do you think about maybe the number of agents over time in the future with that in mind and with maybe AI being part of the labor arbitrage?

Yes, it's a good question, Samad. I'll take that. First and foremost, we're really in the business of providing a software platform to manage all types of interactions, whether they're agent handled or they're automated. So we're really the beneficiaries as long as interactions continue to increase and customer interactions in general across automated versus human. They're going to continue to go up. They've gone up for decades, and they will continue to go up for decades with the economy. So we're in that business. We also -- again, we could debate for everybody is taking different sides of this debate as to how big of an impact on the agent population might occur. Again, we're beneficiaries. We actually get double the revenue per customer, if the agents are replaced by automation with our solutions. So we're beneficiaries. But let's just -- I think we could say that it's going to be no impact, little impact, significant impact. We've actually talked to some of our largest customers. We had a customer advisory board meeting last week. I asked 15 of our largest customers, what they expect their agent population to do over the next 3 to 5 years? None of them said they will significantly decrease their agent population. That said -- and by the way, the analysts like Gartner and IDC and others have also projected agent population not declining significantly. But even if it does, we're the beneficiaries. Again, efficiency and productivity gains accrue to the software providers at the end of the day, especially for providing solutions across that spectrum.

Yes. I think I just want to double-click on that just for the audience clarity. You mentioned getting -- that you get double the economics for an IVAC versus for a seat that a human agent is using. Maybe just help us understand how software spend compares to the overall spend inside of a contact center? Because I think that's a really big part of the opportunity that maybe gets lost in this discussion.

Yes. A big, big cost, like 80% to 90% estimated cost of running a contact center, operating a contact center. 80% to 90% is the labor and 10% to 20% is technology. And so it's a huge labor arbitrage opportunity. This automation opportunity, the ROI that our enterprise customers gain by implementing our virtual agent technology is so significant. They're typically -- if they are going to actually look at this as a replacement as opposed to an augmentation of the agent, it's a 10:1 cost savings in terms of labor costs versus software costs.

Understood. And I don't know which of the three, this is for you, but are you finding the payback or justification from your IVA or other solutions that you've implemented?

So I'm not sure Samad, ask that question again, sorry.

Yes. So when you think about the IVA solutions and the payback or justification for what you've spent there, whether it's Inference or whether you've implemented yourself. Are you finding that payback there?

Oh, absolutely. And I was going to answer it in the context of a customer, but I already have. Our payback has been tremendous. Again, this is an amazing revolution that we're part of that Inference acquisition was very timely 2.5 years ago by us. It put us out in front of this AI and automation opportunity well ahead of our competitors. And frankly, we believe that we're still getting started. So I think we've got 8 AI and automation products already built on top of that Inference IVA solution that we acquired, and it's exciting times.

Understood. A question we get asked often, and it's more -- it's, can the end business. So choose your large brand, can they actually train their own models? And can they build it themselves? How should we think about that versus their reliance upon Five9 to help them with that journey?

Yes, great question. So the answer is that enterprises need a platform that handles all the customer interactions. And that's actually a big part of what's hard to build. If you want a system that can handle chats, e-mails, phone calls, both inbound and outbound, can handle WhatsApp, all these type of different things and bring it into the system and then you need the analytics and reporting and you need the routing and you need to handle cases where it goes to a live agent so you need real estate in front of the agent, UI, UI for the supervisors, all that stuff is what our platform does. And that's really hard to build. And that's why customers instead of trying to build that on their own, they come to vendors like Five9 to provide that. And of course, they want automations using large language models that allow them to improve the efficiency of their contact center and they look to vendors like us to build it because it can be integrated into the platform. Now you talked about customizing for their own use cases by training their own models. With large language models, it's no longer necessary to train the model on your unique data. Training is replaced by prompt engineering. That's now how you customize it for use cases. To do prompt engineering, what you need is a system can go gather all the data about the customer, gather and integrate all the knowledge in the contact center, go integrate customer context, like what were the past calls and chats that they had and what was discussed and all funnel that into this prompt, that's the input to the large language model. To do that, you need a platform that has integrations into all these different data sources and APIs and back-end systems, and that's exactly what we do. So that's why historically, customers have looked to vendors like us to provide these things. And we believe with large language model, there's even more incentive for them -- to vendors like us because that's exactly the integrations that we provide.

Great. So we've talked about the -- what Five9 -- where you sit in the world. We've talked about the technology where AI is layered in. I think at the end of the day, we all -- like the big question is monetization, right? It all boils down to money, brass tacks. How are you thinking about -- how are you monetizing it today? And how are you thinking about the pricing model going forward in an AI-driven world?

Yes. So some of it, we covered already, but to synthesize it all. We have this platform that we talked about. The admission to come on to the platform is $200 per month per live agent currently. And then the main product that we've been selling because it's been around the longest, is the IVA. As Mike said, we go from $200 to $400 for a given set of interactions. We sell that by the [ portal ] replacing the human agent. Now that's just the first of the 8 different products that we have. So it goes up from that $400 when you start implementing other things like summaries and AI insights and workflow automation, et cetera. And we're in the process of more precisely -- Jonathan is in the process of embedding that throughout the contact center software that we sell. And so a big TAM expansion potential from -- as we go up with these other products as well.

Great. And then maybe what are the implications? Let's say, if we move to more interaction-based model, what's the implications for software spend, right? I think -- everybody is trying to do that math, and it's early days, but how are you thinking about it as an organization?

Yes. So we talk to our customers all the time about various pricing models. And again, by exception, we will price virtually any of our products on a usage model. They almost always prefer to go back to the way we've done it. It's very predictable, if we do it on a per seat, per IVA, per module as opposed to on a per usage. And in the end of the day, most of our customers realize once we do the math with them, if it's per usage, they're going to actually end up paying more, and it's something that's not predictable for them. So most large enterprises really -- they've got a budget, they've got an ability to kind of contain that expense to provide this solution, and that's what they usually prefer. But again, we're going to be flexible in go to a usage model, if that's where the market goes. In the end of the day, we'll still monetize at a very, very attractive return for us because the ROI that we're delivering to our customers. These are significant tangible ROIs for almost every single one of these products.

Understood. I know we only have a few minutes left. I think one of the things that gets lost in all of this is just how complex it is to actually architect and design a contact center. I mean if you think about an airline just because I flew here in Virgin Atlantic, a large global airline running 24/7 operations all over the globe. And so if you think about the customers that you're serving and think about how they're building out these complex operations. Maybe what are -- how hard is it to design that? And how does Five9's professional services environment help with that? And why is that a moat for the business?

Yes. Absolute, great question, Samad. So first of all, this is like a heart transplant. This is one of the most mission-critical systems for these large enterprises. It's their front door for their customers to interact with them and transform -- going through that digital transformation off of these large legacy on-premise solutions into Five9 in the cloud. It is like heart surgery, and it requires expertise. It requires a full platform, as we've talked about with the airplane, if you will, to replace those via Genesys, Cisco platforms. And at the end of the day, it requires professional services experts. We've got a team of over 500 people that are in our professional services organization. They mostly come from the legacy on-premise world. They've been with us for years and years. We've grown a large team of some of the best experts in the business and it takes both technology platform and expertise, consultative, technical expertise to be able to basically make that transformation off those legacy systems migrate their entire contact center infrastructure over to us in the cloud. This is a multi-quarter journey in most of these large enterprises, some of our largest enterprises that we've added in the last couple of years are almost 2-year implementations. So it can take quite some time and it requires a lot of expertise along the way. And it's a big, big competitive moat for us to why we win a lot of business against our competitors, quite frankly as our team.

Awesome. Well, in the interest of time, we'll leave it there. But Mike, Barry, Jonathan, thank you so much for joining us and really enjoyed the conversation today.

Thank you.

Thank you.

Thank you. Appreciate it.