Amplitude, Inc. (AMPL) Earnings Call Transcript & Summary

All right. Well, let's get started. Good afternoon, everybody, and welcome to the Morgan Stanley TMT Conference. My name is Lucas Cerisola. I sit on the U.S. software team here at Morgan Stanley. And today, I have the pleasure of hosting Amplitude's CFO, Andrew Casey. Andrew, welcome.

Thank you.

And before we get started, just a few disclosures. For important disclosures, please see the Morgan Stanley research disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representative.

So Andrew, welcome. Amplitude has been an incredible company over the last couple of years evolving over the course of that time. And you've described 2026 as transformative with AI agents and new pricing. So at a high level, what business is Amplitude becoming over the next couple of years?

Well, I think it's -- increasingly Amplitudes becoming that infrastructure layer, that data layer and decision point layer that's bringing together behavioral heuristics, such you continue to see evolution in software and products and then increasingly, we're becoming more and more relevant into the marketing use cases. And I would say the customer experience use cases. What I say all the time is that any customer who has a digital interaction with their client, they're going to want to know how that interaction is happening, how they can provide better services, provide better products and make those interactions more meaningful. The use cases around that are how do you drive greater -- free to paid? How do you drive better conversion associated with the loyalty programs? How do you drive better interactions and repeat purchases with your existing customers? And we have a pretty broad base of customer industries, everything from financial services to retail to fast casual restaurants to highly, highly technology companies like even one of the largest foundational AI models uses Amplitude.

Got it. And when you think about the next 12 to 18 months, what are the two or three key variables that you're most excited about across the entire business?

Well, when I started, we had aspirations to be a platform. And a lot of people gave us not a lot of credit for the ability to go build it. In fact, a lot of people said, "Look, everybody says that." And we also talk about major strategy moving more and more into the enterprise. And a lot of people said, "Well, of course. That's what a lot of companies that start off in SMB and mid-market want to do." But I would say over the last 1.5 years, that's exactly what we've done through both this vision of a platform where all these applications we saw in the ecosystem that we're leveraging analytics to drive outcomes for their clients, whether you think of that in terms of experimentation or -- which is like AB testing or session replays or guidance and surveys. These are all applications and businesses that are $100 million plus that are out in the ecosystem that are surrounding analytics. And customers are having lots of pain points about how they actually administer workflows by ETLing data from one place to another, dealing with UI issues, dealing with data taxonomy, and it wasn't very efficient. And so the vision was to bring all those applications together in one environment, such that you can easily draw those workflows between those applications. And it wasn't easy. I think that if you would have asked me when I first joined, you would have seen us with a bunch of applications that have different software development kits that looked like they were separate. But the engineer team did a great job of bringing that together. And for those people who are not very technological savvy, I tell them, think about a time when we had browsers. And if you click in a browser link and then another window would come up. And then you click on another link and another window would come out. And suddenly, you had a sprawl all over your desktop all these windows, right? And then FireFox came out with this great idea. Let's take and make those tabs into the one environment you're seeing and you get greater organization. Well, that was a great way to think about how you better manage your interaction in those environments. And I think about, foundationally, that's what Amplitude did. They brought all these disparate applications into one environment such that you're not ETLing data back and forth. If you get an insight and you want to run an experiment on it, it's that easy to go just run an experiment. When that experiment comes back, if you want to drive a workflow into targeting a cohort of customers based upon that experiment, you can go enact a guide or a survey on it real quickly. Now Spenser Skates, our CEO, has had this vision that product analytics is going to evolve to the point where you have self-improving products. This notion that products are evolving based upon user interactions real time and that, that becomes increasingly more personalized. Well, the advent of agents into our platform is enabling that to occur because now agents are the ones who are picking up on insights and making recommendations, running experiments and if you allow them, to autonomously start making changes to the environment, whether that's a web application, a mobile application or even a business process. So this notion of self-improving products is not that far away, given some of the changes in the architecture that we drove and the agentic capabilities. And so when you ask what I'm most excited about, it's this notion that we can go drive consolidation into a very fragmented market associated with product analytics, marketing analytics and data workflows is becoming more real. And because we did both the technology and executed well and on the sales side, we did well to modernize our go-to-market and create a more enterprise motion than we had before, those things are helping to reinforce themselves. And so even if you told me the agentic capabilities and all the things that we've introduced into our platform, we're not happening. I'd still tell you, I was excited about the two areas that are driving our growth around platform consolidation and the sales go-to-market team moving more and more towards enterprise.

Got it. Yes, it's really exciting. And the AI agents are a new functionality that you've developed. Could you talk about the early customer feedback and the behavior that you're seeing early on with how they're interacting with analytics and guides and surveys and all the other offerings that you talked about?

So the things that we rolled out a few weeks ago was -- you got to think of it as agentic capabilities that drive optimization and broader use within the platform itself. So our global agent, for instance, if you look at the Amplitude UI now, you see a prompt, not a set of tools and widgets and things you have to go to navigate. It's a natural language prompt and you may say, "Show me the conversion rates in the funneling campaign that I had running last night." You'll get out the information, maybe get a few charts. So it enables the barrier to adoption to be lower and starts enabling the business analysts as opposed to the technology analyst to use Amplitude. Plus you have basic agents around automated insights. And I can turn to an agent to say, "Hey, these are the things I want you to focus on and the data sets that are coming back from the mobile application." And so the agent can run continuously about how those interactions are happening and give you insights on to what the user behavior is. They can also start running experiments on that data continuously based on the insights of the agent. So you have an agent to agent pass and the experiments can start running. And we have an agent that does summarization of session replay information. It's one thing agents do really well, is take large data sets and they summarize them for you. And think about the engineers that had to go watch lots of different session replays to get qualitative feedback on the interaction of the product. Now the agent can summarize it and say, "These are the ones -- don't go look at 1,000 of them. Here are the three that exhibit this behavior that you should look at." And potentially, you might want to run an experiment to understand if you change that environment, what would be that outcome. So those things working together really will exemplify the power of the platform. What our customers have been exposed in the beta have learned is that it's much easier to drive those workflows and much faster to run some of the campaign analysis or promotion analysis or product feedback than they ever could before. So when we started talking about the agents and the optimization, that was the first thing that customers are like, "I want to learn how to do that." And I'll talk about pricing and packaging in a minute, but the first thing they say after that was like, "Wait a minute, this is going to cost me a lot, too. So help me out with how we adopting that is how much it's going to cost me." And we said, "Well, we have a response -- we have a solution for you." The other thing that was interesting, we started talking about MCP server and being able to connect to more data in a more seamless way. And one of the biggest impediments to customer adoption of Amplitude has always been the underlying data taxonomy. Most enterprise customers you go to them and you say, "Show me what your data architectures are and how you want to use that within Amplitude." they'll show you a mess. And the reality is MCP starts to help obviate some of that mess because now you're not having to do hard connections and integrations, you're actually able to use API calls to get through to relevant information. It doesn't help necessarily quality, but the quality can be addressed through an agent actually working on the relevant data. I'll give you a use case where a customer started to bring in their CRM data into Amplitude. And they had 5 different definitions of ARR, which one is right? Well, the agent dissevered that the one that's used in most other places in adjunct applications was top line, and that became the basis of the integration. So there's ways in which we can use the technology to really add more and more value that customers can implement and more and more data that customers can implement in the platform and a lot of them are very interested in that. Now we knew that as we were going through this platform journey, and that -- and this move into enterprise selling, that our pricing and packaging was not right. It was absolutely inhibiting adoption. It was a very complex, it created a friction in the selling process. Customers didn't have transparency on what their usage was going to be. So they are very worried about the cost associated with adopting more of Amplitude. And we had really poor selling practices, too, that had penalties associated with customers going above and beyond their entitlement. So you think about a customer who has a million events as part of their contract. And maybe we're charging them $20 per million of events ingested. We'd have a -- if you look at one of our order forms, it used to have a clause that basically said, "I'll charge you 2x or 3x if you go over that." So talk about dissenting adoption. You're putting the fear that the customer is going to get a bill at the moment that they're not managing their data ingesting correctly. So we flipped that. And we said, "Look, if we're going to be a platform, you've got to -- the first principle is you got an incent adoption" and you've got to lean into the customer value proposition, help them understand how they're going to adopt, be more transparent on what those costs are going to be as they adopt more, give them an incentive to displace other applications that they may have around this ecosystem. So we had to have an approach of the philosophy and our pricing strategy was not a premium pricing. It was one that was value for the money. But if I go into any customer and said, "I'm going to replace three applications", but they pay us is less than what they're paying those other three. So those things had to be abolished. And actually, when I joined Amplitude, I gave the Board a report out after 30 days. And this is one of the things I focused on. I said our pricing strategy is broken. It's actually causing churn and it's not allowing us to go drive our platform strategy nor our enterprise strategy. So we've been working on it for a while to simplify the architecture. And we tested a lot the primary billing meter we have. Unlike other companies, Amplitude was already using a volumetric billing meter with about -- based on the number of events that were ingested into the platform. So it's not seat-based, it's already on a volumetric. We didn't know if that was going to be -- should continue it. But customers came back and told us, no, actually, that data ingestion is a great proxy and value that we're getting from Amplitude. So we kept it. And it's good because it was 86% of our installed base was on event-based volumes. What a lot of customers didn't like was that when they adopted more of the platform, they got additional billing meters. So they adopted experimentation. They had a meter on experiments. They adopted session replay, they had sessions. They adopted web analytics, they had web sessions. And so they said, "Look, that's too complex, and that's pushing more license administration to our clients." So what we need to do is simplify it as much as possible. So now rather than having specific meters, we have a value-based uplift based on the data ingestion. So if they're paying is $20, like the example I gave you before, they may pay us $26 by adopting experimentation or they may pay -- if they go another one, they may pay us $32 per. So it's an uplift based on the data they're using and how they're using across the different modules themselves. And why that's important is because it gives implicity and cost predictability, transparency to our clients, especially if we're going to use this over a number of years and allows us to go right back in and show how customers have that cost predictability when we quote against the installed base competitors.

Got it. And could you talk about the breakout between those within your existing customer base, who've adopted this pricing versus those who haven't? And what do you expect that to grow this year? And then maybe, over the next two?

Yes. I'd tell you that we were testing it through Q4. There were -- one of the things you do when you introduce a new pricing package, you don't try to automate it right away. You try to make sure that it's going to be the right one, you test it. So we did that with a number of clients. And what I can tell you as an example, and I expect more of this will be the case that the customer is talking to you first had aspirations to replace incumbent with a session replay and their analytics. That was -- it was a 2-product solution they we're looking at. We started talking about pricing and packaging, what we are -- and they were expressing interest in experimentation and guide and surveys. We showed them what the pricing and packaging could look like. And they said, "Well, I'd like to do that now." So what happened was we increased the size of that deal. We also increased the contract duration of that deal. And I think they're going to be a great customer for us longer term. So a lot of times, when you simplify things and give cost predictability, they look at it very favorably.

Got it. And then kind of combining both the new pricing and packaging with AI agents, one of the things that we're really excited about is seeing your core customer base expand into more nontechnical users. Have you seen that more recently with new updates? Just talk about...

Yes. I mean one of the -- another one of the customers I was talking to was a pure marketing case, and they were having a lot of pain between what you consider IT or data, the data environment versus what they wanted to run on their campaigns on a regular basis. And they looked at Amplitude is breaking down that barrier. And then quickly, they could manage their campaigns on a regular basis and not have to worry about the data transfer from the IT department. So there's definitely that aspect happening as we move into more and more use cases and the frameworks around agents just enable that business process to work more smoothly. And the thing we learned, too, is I'll go back to pricing packaging, the other thing we learned is we added agentic capabilities. The customers are tending to adjust more data into the platform because they're seeing more use cases. And they have a high propensity to use a broader set of suite of products. So both the cross-sell and the upsell mechanisms are -- and the pricing framework we're applying to it allow us to monetize as people use more of the agentic capabilities.

Got it. So talking about net retention here. You're currently around 105% with long-term goal of 115%. Can you help walk us through how we get to that 115% goal and maybe the time line that you forecast?

So I think you're going to see us continue to be innovative on adding more and more modules. So that cross-sell. The cross-sell is really what's driven us from that 96% to 105% improvement because throughout the last year, 1.5 years, we've had to overcome some overcapacity sales. So even if data ingestion rates were increasing in our core customer base, we had to overcome these contractions and churn that have been caused by that poor selling. I'll give you an example. I had -- one of our largest contracts was renewing in early 2025, and data volume hadn't changed, they were using. But contractually, they had the ability to reduce their rate by 33% upon renewal. So why would we have ever taken the full value of the ARR at the time? It's not something I would have done, but because there's a high propensity for them to renew if you're giving them a 33% reduction and they're using. But that's some of the bad constructs that were out there that we had to overcome. We're past most of those. And so what's good is data rates continuing to increase in the platform, and we don't have those contractions, suddenly, upsell has become a meaningful aspect of net dollar retention, okay? So let's call that 5 points, okay? Now cross-sell continues to get more and more robust. And we still have -- in our investor pack, we showed how -- the percentage of our ARR that's resulting from customers who have 2, 3, 4 and 5 products. We have 74% of our customers are actually on 2 products now. But if you go down to 5 products, there's only 20% of our customers. So there's still a great opportunity just within that product set for us to expand. And I would tell you, we're going to be introducing some new products in the very near future in the marketing analytics space, in the agentic search space that we'll charge for.

Right. And there's a ton of new app development, obviously, with AI facilitating that process. Could you maybe talk about the breakout between -- and you kind of just touched on it a little bit, but the opportunity within your existing installed base versus acquiring net new logos on a go-forward basis?

So I'm one who preaches, especially at this level, we're -- call it, $350 million to $400 million in size. We still should be generating a good proportion of our ARR that's coming from the new logos. And I was worried when I first joined, we were over 30% new logos and 70% expansion. And so we started re-architecting territories, driving prospecting aspects more, our marketing demand gen was targeted more at new enterprises. But enterprise new logos take longer. You got to set up a master services agreement. You got to do a data protection agreement, you got to do security reviews. It just takes longer, especially in enterprise. So that was a little worry too coming in, like how do we start shifting this percentage a little bit more? Well, in Q4, and we saw it in the beginning of last year, too, there was better balance. You saw more new logo ARR and I would love to get it more like 50-50 or 40-60 as opposed to 30-70 in both first quarters, you saw more new logos and that's last Q4. It was a really great balance. We had a record number of new logos in the greater 100,000 new lands, and we had 18 new G2Ks. So I think you still have -- we're still in a place where we have plenty of room to go after new logos. We only have 160 G2Ks in our contract. So there's -- we're way underpenetrated in the Global 2000.

Great. And then as you continue to make more investments going further upmarket and adjusting to this new pricing model, could you talk about long-term gross margin profile and more of the unit economics as you make these transitions?

Yes. So gross margin, we're actually a pretty simple business from a cost to serve perspective. The biggest cost we have is our hosting costs right now. And we're primarily in AWS over time, we'll probably get into multi-cloud. But with multi-cloud, we'll open up more retail, as you can imagine. The other costs are associated with software applications used to help run the environment. Services costs where we're actually doing professional services and that was a straight cost for a long time, we weren't even charging for services. And then you have cost associated inference, which is relatively new, still relatively small. But all those costs are built into our cost to serve. Now one of the first things we did when I joined was we put a plan together, and that plan said that engineering had to take some effort and focus on driving the marginal improvement cost of data into the platform down so that we're constantly re-architecting and reengineering how our application runs in hosted environments allow us to -- as we add more customers, add more data that it's not linear increases for us, right? So we're driving that down. I talked about the services business. We're never going to have a huge services business. It's 1% today, but it could get to 3% or 4% over time as we get more and more complex. And frankly, more and more customers are asking us to help them with their genetic implementations of workflows. But we're also going to build a partner ecosystem to do a lot of that for us. And then we got to make sure that as we're increasingly seeing larger and larger customers move into leveraging our agent capabilities and we're managing those expenses and monetizing appropriately. There was a quarter, I think it was Q3, where we saw very much an increase in the data ingestion in the platform. Well, we've already started making changes in our go-to-market. We weren't going to drop an invoice on our client, we're going to have a conversation with them first. And a lot of times, those conversations came back. Yes, but maybe I need to align it to my budget a little bit better. So the monetization took a little longer but it also is more durable. So some of those things, I think is a push and pull on the gross margins. We made improvement. We improved gross margins year-over-year. And our long-term aspiration certainly being in the low 80s. But there are going to be points in time where we may see spikes in demand in one area. And I would encourage everybody to think about when that happens, we've instrumented it such that we're going to monetize it and keep driving improvements. The same goes for sales and marketing. Sales and marketing is too high as a percentage of revenue. It needs to be in the low 30s. We're in the low 40s. G&A when I joined was 17%. Now it's at 12%. So those two areas are continually going to drive scale. And the R&D is the source of our future innovation. So I want to try and keep that right around 18% to 20% regularly. And I'll come up when we do small acquisitions and go down when we see greater scale.

Got it. And one thing you mentioned in that intro was partnerships. Could you talk about how those partnerships, particularly with the data platforms change the growth algorithm going forward?

Yes. It's funny. It's another -- I told the Board in that first 30 days, like here is an untapped area for us as far as future demand goes. We were only getting 2% of our pipeline from partners. We didn't have a great relationship with AWS, so we had no go-to-market relationship. So we renegotiated that whole contract now we do, and it's much, much more positive. You see us to have stronger relationships with technology partners. They're in different parts of the, I'll call it, path to which you're delivering value to clients, like Klaviyo and Braze would use an analytics layer to go drive what they're doing in their last mile marketing. There were good progress with HubSpot. And there's a number of areas that I think we're increasing to see with like [indiscernible] and Merkle and others in the value-added reseller space. Ultimately, if we create a framework under which they can create value on top of Amplitude, maybe they develop their own agents for logistics or for retail or for health care, and they start monetizing that. That provides a basis under which increasingly more and more global systems integrators would want to do something similar. And so over time, I think that channel has to be something that is north of 30% of our overall pipeline. And as we get more and more embedded with enterprises, that's the way they've historically consumed.

Right. You've talked about in the past winning simple and winning the enterprise. So when I think about app development going forward, there's a lot of new entrants into the space that may not be your key focus area right now as you move upmarket. How do you balance the investments needed to maintain that demand profile from the lower end, but also move up market?

It's a push and pull, you have all the time -- and just yesterday, our executive team was talking exactly that with some programs we're rolling out to some of our VC friends for their portfolio companies who are start-ups. And we're not going to make a lot of money on that group, but we'll make good money once they grow and start developing their own application and value sets. I mean at one point in time, there was a number of our customers who were very small, and they graduated into paid plans. And we see that continually.

Got it. So let's shift gears to competition and positioning. You've highlighted an accuracy advantage in complex agenetic queries versus the warehouse native platforms. Can you speak to the durability of that differentiation? And where do you see that going in the future?

I think that's one of the areas where you just have to continually innovate and provide the context back on the use cases that customers are valuing. If you're going to pin down our engineers, they'd tell you there was a lot of work to give those queries context and understanding how to get to the right level of a query. Which queries the query that follows on when you get the answer, and how best to instrument that. It is, in many respects, the underpinnings of what we call our behavioral graph which is we spent years and years developing. I'm not suggesting that, that gap doesn't close over a period of time or that it doesn't change in its dynamism and complexity, it will. But we'll continue to recognize that that's the case, too and innovate around that.

Got it. And as product analytics and marketing analytics converge, how does your simplified pricing model and AI native architecture meaningfully improve win rates versus the legacy vendors...

One, the pricing and packaging does a great job of that because we can go right into any customer pretty much and say, "Look, if you've got 3 or 5 applications that are competing, we can show you a value for the money where you'll pay less in licensing" by standard on Amplitude. And we can -- we've also done, as I mentioned earlier, a great job of showing how you can optimize the workflows. They used to be very cumbersome across those application environments. And I would tell you from a product and marketing analytics space, every business is trying to figure out how they better digitally engage with their clients, whether it's a service or a product, what have you. They are trying to figure out how they do that in a more efficient and more meaningful way and get better outcomes. And that means that more of the classic marketing cases are moving towards consider treating their campaign as a product or cheating -- creating their promotion as a product. And that lends itself to the architecture that we've built.

Right. Let's shift gears to financials. RPO grew 35% year-over-year and looking at contract duration that's nearing 22 months now. How is the structure of these deals changed for the past year? And how do you expect them to continue to evolve?

Yes. There was -- there used to be a philosophy in our sales team that customers didn't want more than 12-month contracts. And I said that's [indiscernible] especially in the enterprise. Enterprise is -- there's always this push and pull of like, "Do you want to pay for something you're not using so you get software, cloudware or whatever you want to call it?" versus I want cost predictability for the term of the agreement and so that I'm not going to be surprised on cost as I start to adopt. And in the SMB mid-market space, they're very concerned about their cash flow and getting unit economics. And that group, I would agree for the most part, they'll want to have 12-month contracts and they want to make sure that they're getting the right technology and instrumentation, especially if they're trying to improve product market fit in their own product. But the enterprises think differently. Enterprises oftentimes either have a homegrown application or they have a competing application and they're planning for when they're going to transition over, they want cost predictability. And so I've increasingly taught our sales team how to have the right deal construct conversations to meet the customer needs. A lot of times, they used to talk about our technology. What I want them to talk about is what the customer wants to achieve. So the customer tells me, "Look, I really want to move over from the [indiscernible] vendor to you, but I have a problem and that contract goes for the next 5 months." Well, "Let's do a construct that says, you're not going to sign up for just -- in 5 months because that's not realistic. You're going to have a transition where you need to use us as a staging period. You need to test this out." Let's set up a construct that's -- let's call it 3 years. And I won't charge you as much for the first 5 months, and we'll not charge you more in years 2 and 3, right? So that's a simplified construct that we've taught our reps, how to do. And suddenly, they -- wait a week, I just got them into a 3-year contract. Yes, you did because you listen to what was most important to them. A simple one, I always tell our reps is if a customer tells you, you have a budget problem, you know what the first question you should be asking? Are you talking in terms of expense or cash? Because how you respond to that will determine how you should construct. So if they told me I have a budget problem, an expense problem. My budget period doesn't start until January 1. Well giving them better payment terms doesn't help with that problem, right? So what you really want to do is figure out how we can get aligned to their budget as much as possible.

That makes sense. Let's talk about IT budgets. Where does the budget for Amplitude typically originate today, and could you speak to the defensibility of that during an environment where IT budgets may get compressed?

So a lot of times, our initial core use cases are product analytics. So you got the CTO, the product manager, the product developer, those are the classic percentage we're going after. I'd say increasingly, we find ourselves with more broadened marketing analytics, you're going to CMO, a business analyst or you're going into potentially even the CIO environment. And when we dealt with the economist, they were moving their business from pure print to subscription, and they're using Amplitude to understand how users are interacting with content and that influenced what content they created, meaning they want to see when customers are browsing versus actually engaging. They use that information then to go build out what their advertising campaign was and they also use those interactions to understand what customers would value in the subscription packaging. So that transition was one that enabled them to move into the digital world, but then broaden out the discussions on workflows and how Amplitude was being used. And the people that we were selling to were the CIO, the CTO and the Chief Marketing Officer, they were together in that discussion because it was so essential to their transformation. So it kind of depends on what the use case is and who you're talking to. But our view over time is there aren't these silos of operations and data that they're increasingly converging. And so it kind of depends on the business and how they're organized and who's going to be the one who's trying to solve the core problem for the business.

Got it. Let's shift back to product for a second. So if you think about the future investments in the business and the current segments, versus investing in new segments, how do you split forward investment between your existing products and the demand you're seeing there versus new areas you want to explore?

Yes. It's a hard one because -- I mean, that's the level of maturity we've had to adopt even in our planning process. I can tell you, we go set a plan, and that plan has a limited amount of funding for R&D. And then they work with sales and our customers to try to prioritize as best they can. But invariably, you're going to draw a line and say this is where we're going with this. This is as far as we're going to go. And there are some things that you would say are enhancements and improvements in existing applications. Like when we did session replay, we had, had a specific gap versus the incumbents out there around mobile. Mobile, we had to have especially with mobile. So that was a high priority. Well, there was others that on the marketing analytics side around orchestration and merchandising and e-commerce that we had to deprioritize. That didn't mean they fall off completely and you're just having to work through the priority list first. And the thing that we often do is we have reviews, we question that. Our customer feedback, we're seeing demand in areas that are more important than others. And that's the job of management to give back the right choices, right? And I'm not saying it's easy. It's not. I shared today a couple of different times that I'm increasingly believing there's a great opportunity for Amplitude in this whole shift from search engine optimization to agentic optimization. We rolled out a product called AI visibility. We had the most views of Amplitude ever. And for us, it was a way to demonstrate how you can quickly create an application, but if you don't give a context and you don't give it value that customers are actually seeking that -- it doesn't really matter. And there's a lot of companies that got very mad at Spencer essentially by saying, "Oh, look, we created this really quickly. But if you don't have -- you don't have user behavior context, you're really not providing the right level of accuracy back." Now the debate we have internally was, was there really a profit opportunity for us? Who's going to pay for this? If everything is through the search is through the LLM is free? I think, well, I think there's going to be a market in which it's created, how Andrew? Advertising. Well, guess what happened? OpenAI opened up an advertising business. And now every brand is asking us, how can you help me service my brand versus my competitors through agentic search. That's an opportunity for us. And guess what? We're working on a product that does that. So all these things, I think that -- the one thing I would say, maybe for me close on this is tougher companies have evolved over the last 30 years, every time there's a new technology that makes developers more productive. They've sought out new ways to drive software development and value to their clients that automate and improve workflows. We're going to continue to do that, too. In fact, we think that in order to survive in the agentic world, you have to adopt fast, you have to move quickly. You have to deliver products at breakneck pace that delivers value to our clients because the world is increasingly becoming more and more personalized through agentic optimization. If you can't move quickly and drive value, and those are the companies that are going to find themselves disintermediated.

Really, really interesting. And that's all the time we have. Andrew, thank you so much for the time. Appreciate it.

Yes. Thank you.

Thank you, everybody.