It never gets less. All right. I think we're good to go. Really glad to have my panelists here to talk about an issue that is near and dear to my heart today, 'cause I've been following the software industry for coming on three decades, I think now, and watching it get absolutely under attack by the fears that AI is gonna disrupt it and break all of the participants. I find very frustrating. I do think if you are a very thin piece of software, you've got real problems. There's a lot of really great strong software platforms out there that I truly believe this can be a great thing for them.
We wanted to be able to talk today about sort of the real-world implications of what's happening and bring some people up who have some pretty deep and unique ability to address this question. I wanna start with each of you to sort of describe the company that you represent, your backgrounds, and that'll sort of set a foundation for, you know, who we are. Why don't we start with Seth Ravin from Rimini Street first?
Sure. We have thousands of customers around the world, from governments, military, to private industry, serving primarily 10 billion and above in annual revenue type companies. We spend our time working with CIOs, CFOs, COOs to bring down the total cost to serve in their business, labor, systems, processes, and now, of course, agentic AI solutions that help give them a different view than just upgrade, upgrade all their software. We're bringing actual innovation over the top of these systems now at a fraction of the cost, much faster time, and reducing labor. We definitely have a view of all of this core technology that relates to transactional systems that run businesses and governments today.
All right. Let's have David, you now can talk to us about Zeta.
What, you can do it better than me?
Yeah.
Come on.
Yeah.
At Zeta, we are a marketing automation platform that primarily uses a very large pool of first-party data and our own artificial intelligence algorithms that we started programming in 2017 to help very large enterprises. 151% of the Fortune 100 largest companies in the world are our clients today out of our 603 clients globally. We help very large companies to create, maintain, and monetize customers at a substantially lower cost and a substantially higher return on spend. I think the exact statistic, according to Forrester, is we have a 600% return on investment that is spent through our marketing platform to our clients.
Let's size up each of your companies so they understand sort of how big you've gotten to be to give some credibility to, you know, your insights. How big is Rimini now? Top line, bottom line.
We're over $400 million in annual recurring revenue. We've been public since 2017 now.
Rough headcount, maybe.
We're about 2,000 plus people operating in 28 countries.
All right. Same for you.
In the middle of our range this year, we'll do $1.755 billion, almost $400 million in EBITDA. 60%+ of that should drop to Free Cash Flow. This will be our fourth consecutive year of 30%+ compounded growth top line, greater than 50% EBITDA growth, greater than 75% Free Cash Flow growth on that four-year compound.
All right. Let's get into the AI stuff. Your company has, for decades now, helped companies who are afraid, literally afraid to get off of a software platform, in many cases an ERP platform. Talk about why software platforms are as sticky as they are that might make it a challenge to simply, you know, vibe coding a platform to replace it in any reasonable amount of time.
Well, I think as you started off by saying, there is a certain amount of over anxiety that somehow a big ERP platform is going to be replaced 'cause someone wakes up one day, and they're gonna vibe coding in a morning, all 27,000 processes and replace millions and millions of lines of code, that is audited, has to meet standards, you know, with auditors around the world and public company standards. This is not gonna happen. This is going to be a gradual process. Customers, when they get involved with massive systems, these do not change out overnight. In fact, the reason we started the business was we are in the business of keeping these systems running much longer than vendors tended to want to support them.
They wanted people to do upgrades every three-five years of massive size, and we run these things for 20, 30 years, and they're fine. We provide the longevity, and that's because they wanted stability, and we provide better stability. Now with AI, what we're doing is we're building new capability over the top. They're still gonna keep these systems for years to come as we build over the top. Eventually, as we've already declared, ERP software, for example, is dead. It will be replaced by an entirely new paradigm and generation, but it's gonna be a gradual process. This is not tomorrow. These systems are gonna be replaced. That is overanxiety for sure.
Yeah. Dave, why don't you talk about the technical sophistication of what Zeta does?
Well-
You know, we'll come back to the fact that it. You're AI capable as is, but it's real time, it's massive data. Like
Yeah. I mean, just to address your first question and then do it. I agree completely. This sort of narrative that large language models are going to take over for all enterprise software. Really, if you step back a little bit and you think about this, I've been CEO of seven companies over 35 years. I remember when the internet was first launching, and everybody said, "The internet is gonna destroy Walmart. It's gonna destroy JP Morgan Chase. It's gonna destroy Federal Express." In all cases, companies that adopted that technology became juggernauts. Then we had mobile, then we had cloud computing. Remember when cloud computing was gonna disintermediate all software? We were gonna have little apps that ran on top of it. Once again, in all cases, organizations like Zeta who have adopted these technologies have accelerated their growth through that.
You know, it starts with fear because you've got some companies that, as you said, are gonna be thin applications, more workflow management tools. I think they're gonna struggle. I think companies that create intelligence, companies that create return on investment are gonna thrive if they embrace these types of technologies. As it relates to the complexity of our business, I mean, to put it in perspective, we have 552 million active people in our data cloud. Multiply that by the average number of data fields per person, 5,000-7,000 per person. Our first-party tracking pixel sits on trillions of pages of content that is ingesting everything these people do on each one of those pages every moment of every day.
You would have to multiply each one of those by the next number to get to the number of computations that we have to do. I think in some cases it's, you know, between 7,500 and 9,000 computations per millisecond every moment of every day, which sort of tends to start adding up at some point. All of this is to create intelligence on what do people intend to do next. Do they intend to get a new credit card, buy a car, churn off a wireless platform? Make sure that we are helping our very large enterprise clients to get their messaging through a curriculum of content, which is another series of algorithms, which is focusing on the best place to focus on these individuals to either help them create customers or retain the customers they currently have.
All right, Seth.
That's the scary stuff, right? The intent.
Well, to be clear, just obviously you're new to our story. We never sell our data to anybody at any price at any time, and we never share the personally identifiable information, not even with our enterprise clients. Everything is masked with a Zeta ID number to protect the consumer, and you're simply trying to get to the intent of that Zeta ID number, which we know is a deterministic individual, but the clients know as a Zeta ID.
Let's talk about from a different angle. What would you think the moats would be to the business, sort of, that make it a challenge for in an AI-driven world?
Yeah. I think when you look at the, you know, the differentiators between the winners and the losers, I sort of joke at Zeta, we're the disruptor, not the disruptee. We started in artificial intelligence in 2017. We re-architected our entire platform. We launched it in 2021. I believe since then we've been on greater than a 30% compounded growth rate as a company top line, many times that EBITDA and free cash flow, because using that data and that AI drives a substantially higher return on investment. As it relates to moats, we have our data, 552 million opted-in global individuals, plus all of our Fortune 500 customers trust us with their data. We ingest all of their data, match it with our data.
To his earlier point, I don't see Fortune 500 companies giving their first-party data to a large language model anytime in the distant future. You've got 600% return on investment today. Our clients look at us as a revenue center, not a cost center to cut. He's gonna know this as well as anybody, the processes and procedures for working with Fortune 2000 companies, data security, data procurement. You've got their legal, their accounting, their procurement groups, SOC 2 audits for many of our clients every year. All of this is very, very difficult. Back to what we originally talked about, I think the companies that embrace this technology and have proprietary data. By the way, we never feed our data into any large language model ever. We keep it as proprietary to our models.
We have our own proprietary models, which we've worked on for many years, and we have our own AI agent studio where our clients have already built thousands of AI agents inside of their data and their platforms using our studio. You know, last year, our net retention rate as a company was 120%. When you think about that, you tend to start thinking that the business is pretty sticky, but when you're creating these very large returns on investment for your clients, they tend to wanna stick around.
Yeah. Seth, why don't you talk about sort of the moats that you see that sort of preclude the AI engines from simply walking into businesses like yours or other software engines.
Well, again, I think as you said, I mean, data, systems, processes, these are so extensive in organizations. What we're watching is really the use of these new technologies to bridge gaps in the current systems and the current processes. The one thing that's so important is we cannot AI wash everything.
Right.
Right? I mean, we all sit here every day. How many of your software releases have companies called you about with a .ai added to it, and they're telling you it's a whole new release? So much of this is just not true. They're really not AI releases. So everyone is being inundated with this, and it's causing a huge amount of freezing amongst CIOs, CFOs. They don't know what to do. There's so much coming at them. I think what's amazing is a lot of these companies would benefit 70% just by improving their own processes, doing some automation. You don't even have to get to AI before we could have improvements in business.
As I said to Rich all the time, I said, "We just gotta calm everyone down, take a look more at what we're doing structurally, understand more about the technology and why we're able to use AI for better, faster, cheaper operation, and look at the problems we're solving." One of the best things that we can do is we create for customers now checklists. We have processes for determining value of AI in an agent. What kind of agent would you want? Right now, people are kinda spinning around 'cause they don't know where to start with all of this. I think we are going we're on a journey for the next 10-15 years. This is another one. We all lived through the dot-com. Well, not all of you. Some of you are pretty young.
Those of us who lived through dot-com, figuring out what HTML was, as you said, "Oh, wow, what is this?" We went through the cloud days. We've been through these new technology cycles, and we, the ones who really do well, as you said, one, adopt them, but two, they calmly adopt them, and they're often not the first people to do it. If you looked at the statistics, the number of failed AI projects is staggering because people don't know what they're doing yet. They're jumping in because they think that everyone else is jumping in, so we're a little bit
Well, their CEO is making them.
Pathetic right now, right?
The CEO is making them.
Yeah. Well, we sit there with boards of directors who say, "We are a digital-first company. We are cloud first, right? We lean into all of this new technology, and we use AI. And oh, by the way, profits are down and costs are up, so we're gonna have to cut the IT budget by 30%." And the CIOs walk out looking like they were beaten up because how the heck do you do this? I'm supposed to cut costs. At the same time, I'm supposed to adopt technology which costs money. How do I do this? How do I move out of status quo?
This is why I have so much fun with this because we get to help companies kind of figure out, "How am I gonna do this?" Well first you're gonna stop the nonsense you don't have to do, and you stop the noise, right? You're gonna save so much money there, and then you're gonna use that money to really invest in true innovation and get off all these upgrade cycles that don't yield anything. So this is really much more about discipline. It's about following charts and ideas and making good decisions. And you just have to get everyone into it. It's a governance model that's huge for everybody.
I really don't believe they can just replace software things. One of the things that's interesting then, if that's not gonna be part of it, is what you can do with it now. Let me start with David. You've recently partnered with OpenAI, their chief commercial officer, and you put a press release out. Talk about what you can do with, for example, that OpenAI partnership.
Yeah, back to what we were talking about, embrace, right? I mean, if you think about it, the vast majority of the investment into artificial intelligence to date is in foundational large language models. It's not where the market is going. It's really going to applied AI, which is gonna be very small models that are gonna sit on top of the large language models. You don't need the entire source code from the internet to figure out if you're gonna buy a plane ticket, right? It's great if you're gonna write a poem, but it's not great if you're gonna do that. What's gonna happen is you're gonna have models built on top of models, and I think that's gonna be what really changes the game on the internet.
If you think about our relationship with OpenAI, which is pretty big today, and we're working on much, much bigger stuff, which is fun. We now have a joint engineering team with OpenAI, where we have Zeta engineers and OpenAI engineers in the same room working on really cool projects together, which is pretty cool. The outputs are incredible. We brought Nate Johannesson, who would really help to build all of Meta's AI and ran, I believe, AI at Instagram formerly before he joined us, and he's been all over this. It's been really cool. When you think about it, Athena, which is to us the future of our business. Not to get too out there, but humans have been around for about 400,000 years. Up until 50 years ago.
We exclusively communicated through voice. About 50 years ago, we created this thing called the keyboard, which effectively everybody hates. It's cumbersome, it's very difficult to communicate, and it creates a massive barrier between humans and technology. Athena, which is our voice-enabled super agent, is a voice platform to help our clients navigate our entire UI. If you think about it, all of us, we build 747s. I always joke with investors, what percentage of your Bloomberg terminal do you use on a daily basis, right? Today, our clients know how to fly a Cessna. With that Cessna, they're getting to a 600% return on investment.
Athena, instead of having to go navigate all of these different things, you'll be able to say to Athena, "Athena, I'd like to add two million incremental customers this quarter at a average of a 7% lower cost per customer creation." She will then not just tell you how to do it, she will show you by changing the user interface and give you all the answers. You'll be able to say, "Great. Please activate to that, Athena." Our partnership with OpenAI is in the building of Athena. We're not exposing any of our data to OpenAI, and we're not using any of OpenAI's foundational models in figuring out intent-based scores or where to target people.
The beauty of using the OpenAI platform for the voice enablement of Athena is she's gonna get smarter about what our clients want, what their needs are, and it's gonna not just help our clients. The last voice-enabled interface we rolled out was called Zoe, and I'll very lovingly say she was clunky, probably being polite. Clients who adopted Zoe spent 250% more on our platform, and Athena is awesome. The hope is that clients will adopt it, and this will be the beginning of a very big partnership with OpenAI for Zeta.
Right. Sort of similar thing. I think the adoption of AI can help my companies, in many cases, create new revenue streams, new features, address larger markets. In your case, your company's done a partnership with ServiceNow, one of the premier software players. Why don't you talk about what that does and why AI was important to you being able to create a partnership like that?
Sure. As you know, Bill McDermott, their CEO, and I competed fiercely for 16 years.
Mm-hmm.
He told me, "You know, you guys were a nightmare for me. Every morning I woke up and so and so was moving to Rimini Street from SAP's maintenance program." We competed so fiercely over the 16 years; at least we developed a very strong reputation. What Bill McDermott realized was that we're developing AI toolkits, and ServiceNow has one, Palantir has one, Microsoft. Everyone's developing an AI toolkit. The problem is people don't wanna buy toolkits. If your dishwasher's broken and they bring you a toolkit, that's nice, but it doesn't solve the problem. It doesn't fix the washer. People want solutions. The orientation of just buying yet another platform itself is a challenge. They don't have the money for it. No one's got the interest.
Bill saw a massive opportunity inside of ERP and transaction systems because no one was really focusing on how to take those transactions and move them into an agentic world where we can lower the cost, and we can make them more flexible and then use agents to run them instead of people, lowering labor costs. That's where we partnered up because he realized they've got the toolkit. We've got the experience and know-how to take the toolkit and use it on these big transaction systems. Bill and I are really partnered in this massive program together. We also, of course, work with Microsoft, and we've got Palantir and others because they all have toolkits too. Everybody wants to put their toolkit to work because we just can't sell platforms. I think this is a really exciting time.
As you said, we believe eventually we will replace piece by piece. If anyone thinks this is less than a 10-15-year journey, right, they don't really understand the magnitude. As you said, we got auditors, we got finance, we got procurement, we have SOX and SEC all over the world. This is going to take time. It is super exciting because we can now fix a problem that in an ERP program. Let me just give you an example. If you were running an SAP ERP program, to make a change could take months because we not only have to design the change, we have to make the change, test the change, go through audit functions. We have to do all these things. Now, instead, we put a Band-Aid over the top.
We released our first 20 Band-Aids for things that are real problems within these systems. They can buy this agentic piece of ERP that goes on top. It's a Band-Aid, and we can deploy it in a matter of weeks instead of months or even years required for an upgrade. We're changing the dynamic, so it is an exciting time. I mean, I've never had more fun since those days because we're really out there changing the world and costs and doing things differently. People need to stop over expectations somehow. All this stuff just gets vibe coded in a morning.
Yeah
deployed, and all of a sudden, we don't have an ERP.
No.
Long term, companies like SAP and Oracle in their ERP products, the big challenge is they're being broken apart into pieces. Everything we're doing in the world of enterprise software now is getting broken into pieces, and then those pieces are getting competitors. So the big companies are losing a lot of their sway within their customers, and they're losing their capability to control. You're watching that happen before your eyes. That will affect, as we said, long-range valuations 'cause nobody can kinda figure out what their revenue streams might look like.
That brings me into a perfect segue, I think, for what I wanted to ask. You both also run companies. You buy your own software or pieces from other people. This can be a threat to certain pockets of the software industry. Why don't you start, David, maybe. Where do you think it is an actual threat to if it's not the big complex platform?
Yeah. I think it's, once again, you said it when you opened the speech, it's thin application sets, right? I think a lot of businesses have been built over the years that are really just workflow management tools. Often, a workflow management tool gets built, it becomes a small business, and a larger business buys it and wraps it into a bigger part of their overall solution, whether it's ERP or marketing or whatever you're working on. What's happened over the last number of years is you have a lot of standalone workflow management tools that have stood out there and stayed out there, and they don't own any proprietary data. They're not deeply integrated into the infrastructure of their clients. They don't really create intelligence.
I think that those companies are going to be disintermediated by, first and foremost, new businesses that are started with vibe coding, where, you know, six people can build that type of a product very, very quickly. I mean, if you look at Claude's coding platform, like it's pretty incredible how fast it can copy that type of software. It can't copy complex software. It can't build internal processes. It cannot deeply integrate into a client that is not gonna share its data with a new third party. Companies that don't have that type of integration point, my opinion, I mean, you might have a different opinion. My opinion is they're gonna have to pivot.
They're gonna have to figure out where do they go from here and/or do they get bought by other companies who are looking at them as a product that can be integrated into what they're doing, and they're gonna get bought for, you know, a lot less money this week than they would've been bought, you know, three months ago.
What do you think, Seth? Who's most at risk?
I think exactly as you said. I use a sliding scale of complexity and mission criticality, right?
Right.
The more you move to one end, the less you're really threatened by AI instead of thinking of it as a way to improve the tool itself or the software. Those on the low end are just gonna get killed.
Yes.
I mean, no one's gonna buy. I mean, they'll get bought up for liquidation value and used, but that's just always happened in our technology. It is the way it is. There are companies you're seeing today that I'm convinced two years from now, you won't even remember their name, because they lost the battle. We are in a pitched battle right now for supremacy and survivability. Who's gonna be the key platforms? We all know it's always a top three, right? In our world, it's always three companies, and then eventually becomes two. We're gonna watch that battle play out before our eyes. The real critical part as you analyze and try to understand what's going on is really understand themes and understand that the players may change.
When you understand the layers and you understand that we think of here is the current software levels, we are now building above them. We need an AI platform, right? Think of it as a multilayer cake, right? Bottom layer is the current software. The mid layer is a piece of AI platform. There's a data layer that may come in, and then there's an actual UX layer. We're right now what we're doing is we're putting that UX layer as stable, so we run our own UX layer, and we can change out any of the AI platforms below it because we wanted to design an architecture. We don't know who's gonna win the game. We have customers who say, "You know what? I wanna use this particular platform.
I wanna use that one. All of the structures we're thinking about now are designing with the idea that there are interchangeable parts because we don't know who the winner's gonna be, and we need to be able to change those parts out while not disturbing what's going on in the day-to-day process. It's really interesting. Once you understand the layers and you really understand what's going on and where everyone's playing in these layers, then you can look at the players in each of the layers and make an assumption some will win, some won't win, and you build architectures that can withstand that.
There is really much to gain from the new technology, but this whole idea that it'll suddenly take out all the big players and they're not gonna have any software to sell. Look, I think in years that could happen, but not in this decade.
I hope I'm retired by then.
I actually I don't think it'll happen in years. I think these large language models are going to become foundational to business, and we're gonna pay them for that. They're gonna make real money, and that's okay because they will make our businesses, enterprise software businesses, more efficient. We'll have fewer employees. We'll probably continue to grow pretty rapidly. I'm not gonna say we're gonna grow faster than a four-year compounded growth rate of 30% at this point, but there's opportunities that come from that. We were, I think, if not the first, one of the first enterprise applications to be put on top of OpenAI, and I can't even begin to tell you how great they are.
We love the OpenAI team, but we use Anthropic, and we use Git, and we use all the Microsoft tools, and we're using Claude, and we're focused on productivity from our engineers, which we've increased by 125% in the last 12 months. We told our engineers a year ago, "If you don't adopt this, then we're gonna find people who will," and we have done that. We've cycled people out who wouldn't use it, and we've brought in people who are experts on using it, and we'll continue to invest in that, and we'll continue to grow as a business. I just don't think that the large language models are gonna end up displacing enterprise software.
You know, you can't say never, but I think for a very, very long time, it's as likely as the internet displacing enterprise software. To remind you, that's what they were saying years ago. Like, it sounds sort of nonsensical now to anybody who has any technical understanding of how the internet operates and how enterprise software operates. We'll look back on this as a time that there was a lot of fear, and there was a lot of new technology. It's very exciting. I totally agree. We're big fans of this technology. Yes, I know I sound like a total geek, but I find it very exciting. I think it's gonna continue, you know, to be a driver to businesses like ours.
I think there was also a moment in time that people talked about open source, that everyone would create software in their basement and give it away for free as a hobby, and no one would sell software. That was probably a decade ago, and that didn't play out extraordinarily well.
Yeah, the amazing part is open source is actually still picking up great momentum. Database, we do a ton of all over the world, Fortune 500s moving off expensive Oracle database, moving to open source. We're definitely seeing trends. Some of these trends we predicted 30 years ago they would fly. They've taken longer because people had to get comfortable with the adoption. They had to get comfortable with the security. I mean, you have to remember, security is the number two question about anything we do today, right? Does it come from China is number one, right? Number two is security.
That's also part of number two, right?
Exactly.
Number one is part of two.
Exactly. If it's number one, stop. You know, if you pass number one, you go to number two and question that. It is. We are living in a world where there are changes. I think the velocity of change is some of that question, and I think we've seen the market respond to a velocity that's not realistic, right? That's the fear side of things.
My only comment when you talk about open source, and we use a lot of open sources, it seems like Oracle and Microsoft are hanging in there. I mean, maybe I missed something at like-
Well, 60% of database-
$4 trillion market cap and
Oh, I know.
half a trillion.
Let's not forget Oracle stock from $350 down to $130.
Well, that's more.
That's a pretty hard jump.
That's more because of their Open AI deal.
That's right.
I'm not saying anything against open source, but we use open source, so we use it all the time. As a technologist, I think it's funny. You will often spend more money on integrators getting your open source to work than if you had bought the actual closed source product. Once again, in the long run, it'll be a lower cost of usage for sure.
I do think that you have to look at these trends, and if you look at the trends, you can actually see where we're going. It's not so much a surprise. You really are measuring velocity. How fast can these changes happen? I think you're just watching the market. The market always moves ahead of where the world is, and the market is looking at it saying, "Hey, you know what? Some of these stocks are a little overvalued. We need to pull back 'cause we don't really understand what their revenues are gonna look like because the world is in flux." I think that this is fine, and I think we're really. Once we get people to understand the technology better, because right now people are just running around.
They don't really understand, especially, I'll tell you, at the CIO level, they're terrified, right? I mean, every CIO, and I feel like I spend more of my time in psychiatric care for CIOs and CFOs because, look, everyone, the biggest fear they have is this imposter fear. They're so afraid that they don't think that they understand this stuff, but they think everyone else does. I walk in, I say, "Listen, let me make sure you understand. Nobody understands this stuff. We're all figuring this out literally as a ticker tape by the minute as new technology rolls out. So don't feel bad asking questions or saying, 'I really don't understand what the heck all this LLM is. What am I doing here? What am I buying? What are the costs?'" It's very complex. You and I were talking earlier.
The pricing models are one of the biggest challenges we're having, right? Everyone's running around, and you wanna ask to say, "Look, if I'm gonna replace a person with an agent or I'm gonna reduce labor costs, I know what Bob or Jane costs 'cause I know their annual salary. I can calculate that, and I know what it is. I come over here, and you say you wanna put an agent on it, you wanna do this. Well, what's the cost of that? You say you're gonna save me money. Well, depends on their skills and maybe if they're, it requires more, it's more tokens. There's a model that's gonna change. We used to do this in the cellphone days. For those of us old enough to remember, you had to pick your long distance plan, which zone you wanted, right? We had all these questions.
that eventually became till someone offered an all you can eat, unlimited plan.
It was AT&T.
Right?
It was AT&T.
Yeah. We saw it happen to their long distance revenue, and the whole thing changed the model. We're moving very quickly in this phase, but you're gonna watch it. The cost models are very complex for business, and that's part of the challenge of adoption right now. Once those become simpler and I can understand what this agent costs and how much it's gonna cost me, a lot of these companies, very hard to tell right now. We're in the early stages, no doubt.
All right. I personally don't think they can replace software companies. Now I think they can help them create more revenue opportunities. Let's talk about internal, 'cause you both now run companies too. Internally, what can you use it for? Do you think it can save costs? We've talked about where your headcount could be.
Yep.
In a year versus today, even as a growth company. You start it.
Yeah. I mean, I think we'll grow the business 30%+ this year and finish the year with fewer employees than we started the year with. When we started our Athena workflow, we started another work group that was solely focused on internal adoption and productivity. We started with all the Microsoft tools. We started with Anthropic. We brought Claude in. We work with Gemini, and we're literally tracking the time utilization that our employees are using them. If they're not, we are training them and talking to them to make sure they do it. I mean, I could see a time, you know, years from now. I don't think we'll grow headcount for years. I think we could be.
Outside of M&A, we could be, from an organic perspective, flat from a headcount perspective, and I think we'll continue to grow our business, you know, on an organic basis at greater than 20% a year for many, many years to come.
We'll definitely be adding headcount, expect thousands. Again, the service that we're providing, we're not going to replace with AI. What we're doing is because we're mission-critical, and if we make decisions that are wrong, people's lives could be at stake. We have human in the loop. We're gonna keep humans in the loop. We use AI to make our engineers smarter. What the AI provides solutions and options that they may not have thought about, so we use that to make them more efficient. We've also used AI tools now to improve the cycle time of response. Now we have probably the fastest response time in the industry. We're 70 seconds to respond anywhere in the world 24/7 with an engineer. We require that because of military and other capabilities.
That is all enabled by systems and processes that are intelligent, and we've been able to reduce our time to resolve an issue from cycle time 28%, using advanced AI tools. We're gonna continue to aggressively roll them out. We did manage to lower our headcount by the end of 2025, lower than it was at 2024. We'll still add, but we'll add at a lower rate because we're using more leverage and technology.
It's been touched on a few times here, but I think it's important. People don't know which AI engine sometimes to pick. I have a feeling in the long run there'll be multiple vendors and that everyone will use across their systems. They'll use different strengths of AI and maybe a low-end solution for one thing. As long as you make it so that it's modular, is it inevitable that it essentially has to commoditize, or do you think that 'cause the valuations on the company suggest that there's never a commoditization future for them.
Well, a lot.
I kinda disagree with that.
A lot of companies have achieved those multiples. They don't always maintain them, right? It's, I think that large language models, you're gonna have a few major winners, but I think you're gonna start to see verticalized large language models. What nobody's talking about is you could use Claude to go build your own large language model too. Nobody's bringing that up, right? If you wanna go build a large language model that's focused on automotive or focused on some different type of vertical, you will be in a position to do that. It actually in many ways will be easier to replicate large language models than it will be to replicate enterprise-grade systems, because you don't really need access to anything but the content on the internet. Listen, I'm a big fan of all the platforms.
We work with all of them. I do think that you're gonna see a couple of big winners, but I think outside of those big winners, you're gonna see a big commoditization to large language models because all of us are looking at it, how can you pull out one model if it's not working and put another model in?
I agree. The world has always gravitated to two or three winners as a market matures.
Yep.
Again, just recognize where we are. We're not in the middle of the game. We're literally at the beginning of the game. I think that's the other mistake people make, is they think we're farther down this road of maturity than we really are. That means that we're gonna be switching out. You've got volatility. Players will come and go. Players will be ahead one day and down the next and could disappear in three days. I do think that if you understand where we are at the early part of this cycle, you can better understand the volatility that's going on. But if you understand the modeling, the large language models, look, that's gonna be a whole segment unto itself.
That's our ability to teach and have agents do different things, and we're getting better and better at understanding that. Now, I do believe that most companies. We're probably gonna use three different sets of agentic AI in ours. We use Microsoft Copilot for all of our back end, of course, 'cause so many people. Microsoft's got a winning seat. I mean, they own the back end of just about every business out there, so they've got a guaranteed seat in the game. Their transport capability to talk to other agents has already been adopted by many, right? 'Cause it's a Microsoft standard. We use ServiceNow on all of our back-end systems we're on. We've got Salesforce for the sales and marketing side. So we'll probably utilize agents across those three platforms.
I think most companies will use three-four different components depending on their needs at that time.
I agree. Sort of the broad-based gold standard a lot of software companies have been measured against for years now is this Rule of 40 concept. In a post-AI adoption world, I think that very much has to change. If you had to generically think about that, where do you think that number goes?
Well, I think we're at the Rule of 52 or 55. I forget. It's well above the Rule of 40, and we have been for quite some time. I think that you know, listen, I remember a few years ago when people were happy to be at the Rule of 40 while they lost money, right? Now it's really, can you be at the Rule of 40 and make money, or can you be at the Rule of 50? I mean, guys like Palantir at the Rule of 70 or 80 just, you know, destroying it.
I would tell you that when you look at a healthy business, you know, we're very proud of the fact we've been above the Rule of 50 now for a number of quarters, but we're more proud of the fact it's a balanced Rule of 50. It's about 30% growth and about a 20% operating margin. It's not 50% growth and no operating margin. By the way, it's not 50% operating margin, which would probably be nice, and zero growth, which would suck. I mean, I think that the metrics by which you define a company change as the times change. The things that never change is what is your retention rate with your customers? Last year, our net retention rate was 120%. That means our clients like us. What's your growth rate as a company?
Are you growing? Are you growing profitably? Are you generating Free Cash Flow? Those are all mission-critical metrics to any business that you operate. I don't think any of that's gonna change. You know, sort of whether the Rule of 40 becomes the Rule of 50 or the Rule of 30, you know, I think it's gonna depend on the week, right, Rich?
What do you think, Seth?
Well, I'm a Warren Buffett type of guy, so I believe in highly profitable businesses that throw off a lot of cash. I hope to get to Rule of 50. That's ultimately.
Well, by the way, we're highly profitable and throw off a lot of cash, so
It would be, you know.
Just to be clear.
It'd be wonderful to get to that number. Our goal is to get to Rule of 40. You know, we're lucky to be in a position that we're trying to figure out capital returns for our shareholders. Being the second largest shareholder of the company, I am very focused on delivering value to shareholders. That is always number one for a profitable business. We're gonna focus on delivering those profits. Adjusted EBITDA is really our goal and our North Star is to deliver on that. Of course, earnings per share. We're gonna become very focused on earnings per share, and there's multiple ways to get there. Obviously, we do stock buybacks and paying down. We're cash positive.
We'll continue to look for ways to build a super strong balance sheet with lots of excess cash that we can put to work for shareholders.
Maybe talk a bit about how you balance or who should drive the decisions around AI internally. Is it? Do you put an AI czar into the company that says, you know, it's their responsibility to drive either a cost side or a revenue side? How do you focus around making it execute well?
I think you just create friction if you take one person and put them in charge. I mean, listen, our CTO and our Chief Data Officer are the best in the world, in my opinion. I think they are amongst our many secret weapons, and I think they're both incredible. I think AI is gonna be pushed down organizationally to every layer of the organization, and part of that is education, part of that is incentive, part of that is tracking. Because if you can't measure it, you cannot manage it. I think that it's sort of a philosophical thought process. The other thing that we haven't talked about, which I think ultimately is very important as you think about your human capital, you have to make your employees feel safe.
If you don't make your employees feel safe using this, they're gonna fear that they're using it to disaggregate or disintermediate themselves. You know, this goes back. I've been doing this for 35 years. Run seven companies, sold four, taken two public. I'm chairman of the other. I would say that. It started to me with understanding how to make employees feel safe failing. How do you make somebody feel like you can fail? Because if you don't let them feel safe doing that, they're never gonna try anything new. It's now I see that in technology adoption. We sort of have. I think we've got a really good culture around you can try things. If they don't work, just fail fast. Don't keep trying to do it if it isn't working.
As it relates to technology adoption, if you don't make people feel like if you learn how to do this and you can replace yourself, you're gonna get promoted. You're not gonna get pushed out. I think that's a big part of adoption.
Well, we have been recommending to our clients and obviously trying to drink our own champagne doing the same thing, but we've really advocated an innovation team separate from our IT organization. We have an innovation team where we can do that fail fast.
Yeah.
We had to do that because we believe that these are different skill sets than running the business day-to-day in IT, and we wanted to isolate that, focus on it, and then have those people work on very specific projects to gain traction. Once they find them, we'll go ahead and adopt them through the company, the ones that make sense. I think that's been on the internal side. The external side, we use a different combination. We have an innovation product, a pillar. We have a leader. We have a head of product marketing for that pillar. We have a head of our, for example, Innovation Factory. The three of them work together to create a triumvirate in the product side of the world.
We do keep it separate, but they do share information with the internal team because we never wanna reinvent the wheel. If we find something great, they work very closely together to keep an eye on the ability to cross-use information technology.
I think the last thing I want to go over a little bit is one of the most difficult and expensive but key determiners of, you know, of a growth company's value is your ability to sell and grow the top line. Maybe both of you, if you could take a moment to talk about how much AI could influence your ability to execute on sales, as these tools sort of come.
I mean, our entire business is built on AI as foundational. We continue to believe that companies that try to put it as a sidestep. I know you have your band-aid, it's different in our world. 'Cause in our world, you need to make a decision in a millisecond. You really can't do that because you're stepping out of a platform to query an algorithm. The algorithm has to hit a data repository. It's got to go back to the algorithm to create intelligence and then inform the platform to do something. That latency tends to destroy return on investment in our world, which is different. When you think about AI as foundational, it is a mission-critical component of our sales process, right?
Today, as I've said, for every dollar a client spends on our platform, we return them between $6 and $7 in business. That's all driven by AI. When you think about our sales motion and when we start with our different clients, we are literally putting AI first as a part of the solution, and we're bringing it in to prove out the return on investment. I mean, we helped the largest wireless company, or I should say one of the largest wireless companies in the United States, lower their churn rate by 37%. That created $2 billion in incremental EBITDA. We did that with all AI, and that was a solution set that really was 10-20 to 1 versus what they paid us. When you think about it, I think he's totally right.
There are organizations that are using AI as part of their pitch, and there are organizations that really have it and are using it. For us, the big differential, I used to joke, we'd walk into a room on a pitch, we'd be the tenth person in, and we'd be the tenth purchase person pitching exactly what we do, and the only person who can actually do it. I used to say to them, "Test it. If I can't effectively lower your cost to create a customer by at least 50%, I'll give you your money back." That was my pitch for many, many years, and it really worked because we could prove it.
I would say again in our world, because again, we're doing everything down to military intelligence, we have to be very careful. We don't introduce new technology lightly. We have huge responsibilities to make sure that what we use works effectively. For us, the fascinating part is our revenues from AI are actually pretty small, but the impact is very big because people wanna know that you have a plan and you're able to adapt. If you can show them that you're able to adapt into the future, they're willing to take the first steps today. When you think about companies and you think about how AI impacts them, you're already using AI.
There are companies, as you said, the promise of AI, the promise of directional capabilities will move people in a certain direction today ahead of the revenues that they're going to generate. You need to look at the influence of these models and where we're going and the vision, especially in the software companies, where they say they're gonna go to, how companies are gonna use the technology, and then you can still actually generate more revenues today on other products while you're getting to generate revenues in the long run on the AI side.
Totally agree.
Right. We're gonna wrap it up. I just wanna say I really agree with your point about the complexity, and the criticality are very important differentiators for, I think, who will win and lose. David's willingness to say in front of a crowd he can grow really fast without adding headcount for maybe years at a time is a very powerful statement about how important this could be to the P&Ls of these companies who do survive because they are mission-critical. While Wall Street has left a lot of these companies aside and really damaged their valuation short term, I think this could be one of the best, you know, developments for the good software companies that's come along in a long, long time.
Hopefully on the other side of this, we'll see a lot of good returns for investors who are willing to, you know, separate the wheat from the chaff.
Thank you, Rich.
All right. Thanks, guys.
Thank you.