Klaviyo, Inc. (KVYO) Earnings Call Transcript & Summary

All right. Excellent. Thank you everyone for joining us. Keith Weiss, I run the U.S. software franchise here at Morgan Stanley, filling in for Elizabeth Porter, who had a beautiful baby girl. So excuse the substitution. But really pleased to be able to talk about a really exciting story, one that's been putting up really good fundamentals, not getting appreciated enough for it. I'll just tell you guys, in Klaviyo. And we have the full cast. We have Andrew Bialecki, Co-Founder and CEO; we have Amanda Whalen, CFO; and Chano Fernandez, co-CEO. So thank you all for joining us.

So 2.5 years since going public, and I think -- thus far, Klaviyo has just proven some of the initial bear cases in terms of too limited of the scope, too limited market opportunity. And maybe to phrase it as a question rather than just a statement. How have you done that? Like how have you guys been able to really sustain the durability of growth through an environment that's been mixed at best?

Yes. Thanks, Keith. Look, we -- since we're, Klaviyo is now 14 years old, I think. And since we started, we just had this thesis that if you think about businesses and organizations on 1 side, and then end consumers, customers on the other. We're just going through this massive shift of, hey, it's not humans in the middle, sales reps, customer service reps, other folks. It's going to be all software and increasingly, we think, just powered by intelligent software, thus AI and agents. And yes, we've been just following that trajectory for the last 14 years. And so that's what's driven us over 190,000 customers $1.2 billion in revenue last year, growing really nicely. But I think what's exciting for us now as we think about what -- if you look at like AI and LLMs and what those have capabilities have unlocked, we've kind of thought about clearly the first 14 years, we built all this infrastructure to basically allow businesses to program the rules, store all their customer data, communicate with customers across a wide variety of channels, for marketing use cases, for service use cases, all of it. But the big unlock was like making that not just things that humans could program the rules, but that AI could write the rules and then optimize it, that's what we wanted to get to. And this is like, I think now -- I think it sounds like it's gets another 14 years. I think use the next couple of years, maybe the next 14 quarters are like the most interesting because I think every business now is going to say, hey, when it comes to how I treat my customers across all services, marketing, sales, service, et cetera, and we're focused on consumer businesses. I think everybody is looking at what is the agentic way or we say the autonomous way to deliver those customer experiences, to have AI that can define the rules of how I treat folks and then deliver those experiences and then learn from them and optimize. So that's what we've been after. And obviously, we've increased the breadth of our products, and we think about like kind of the infrastructure that's underneath it. And then back in September, we launched our 2 agents for marketing and for customer service. We've seen really good adoption there. And yes, we're excited to -- the future is autonomous customer experiences, and we're excited to usher that into hundreds of thousands of business today and hopefully millions in the near future.

Excellent. I mean 1 of the things that's always been impressive and compelling about the story from the get-go is that closed-loop nature, right, that it's not just automating a process for customers is automating a process for customers, letting them understand the results, feeding that back into the system to further automate the process. And when we think about that in context of AI and these learning systems, it seems to be a really ripe opportunity for exactly the system that you put together. That's actually a question for Chano. So Chano, you've had a really impressive career. Co-CEO at Workday, leading April to AI through a lot of AI-driven growth. You're on the Klaviyo Board. I'm sure there's something that you saw of how you can make the story better. You're not going to join the company unless you see an opportunity to make it better. What was it that you saw on Klaviyo? And how are you going to take what's already a good story that's operating really well, how are you going to make it better?

Yes. Thank you, Keith, and thank you for having us and everyone for joining. I mean, well, first and foremost, I would say, because I had to put it in to be as part of the Board, and I clearly saw that there is a shift coming, and I think I don't need to tell all of you what is the shift about. I really thought that Klaviyo was really built for that future and for the shift, particularly for that infrastructure and that real-time customer engagement opportunities that it provides. And then that was the time to do it, for sure, right? Obviously, it's about people. And I think the world of Andrew is -- has a really great vision for the future and quite of a unique founder. And I think we can do amazing things together. So it's very exciting about that. And last but not least, Keith, I spend my career mostly working with large enterprise companies. We are ready. We know how the infrastructure, we have the foundation to move towards the enterprise and building that market opportunity, which is certainly even larger than the 1 we've been navigating all. So that's really exciting.

Got it. So it's really the intersection of, one, an expansion in sort of the solution portfolio and what you're focusing on. So going from more of just a focus on marketing lens to being the autonomous B2C CRM. And at the same time, expanding out the customer focus of it's not an SMB solution. It's 1 that's now ready to move upmarket into the mid-market and into the enterprise. And those 2 axes enabling a very nice scaling function. Maybe to start on the TAM and the opportunity side of the equation, what does autonomous B2C CRM mean, right? Can you walk us through kind of the vision and how the agentic layer is going to further that vision and enable it to really work for the end customer?

Yes. Well, so here's how we look at our product portfolio. I think there's -- when we think about like traditionally pre this wave of AI, how we thought about software, it's either like infrastructure, we thought about that as primarily stuff you sell to developers and developers used to build applications. There's applications and then the applications kind of get muddled because they're like, wow, some of it's kind of like domain specific infrastructure and some of it is just like a nice polished UI. What's great about LLM is, I think they've expanded the scope of what software can be. And now it's like, hey, look, all these like human users that are running around using software as tools like that now -- that can be done by artificial intelligence. And that's just a huge new opportunity. So for us, we look at that and say, hey, for every dollar that's spent on Klaviyo, there's like between $5 and $10 businesses they're spending on like human capital to go run it. And not -- I mean, frankly, particularly efficiently, right? We talked a lot of folks are like, hey, do you think you really know -- do you have a good sense of like what's your customer experience, like whether it's on customer service or marketing should be. No, I don't, you guys have best practices. Can you help coaching up on that? Hey, you feel like you're getting to all of your ideas. The answer is like, wow, I'm getting more done, but no, I mean still there's stuff that I don't know or I can't get to. So we look at that and think, man, so you've got a whole bunch of people that are doing things. It's actually not optimized. They actually could be driving more customer engagement. Well, great. We're going to build agents to go do that. So that's 1 big opportunity. So we have a whole product group that is effectively what we call our autonomy group that is just focused on building these agents that will sit on top of software as infrastructure. And then underneath that, we think about like, hey, what is the best infrastructure, best platforms, best software that agents are going to prefer in use when it comes to storing all of their customer data, indexing it, allowing making it acquirable in real time than actually using that to communicate so messaging, but messaging not just for marketing, but having conversations with customers, whether it's over webchat or voice or through text messaging, and then what -- how are they going to define the rules and running experiments. One of the really cool things we've seen now with our customer agent product is now that people have this always on agent that is like good at adhering to the rules of the business, policy of the business, we're seeing these like CX teams and operational teams, they're starting to experiment with the policies they use to run their business. Like how do they treat customers. For instance, who gets a refund and when. And all of a sudden that used to be something, man, it's so hard to get adherence with that across like a customer experience organization. Well, agents don't have that problem. And now we can start to experiment and say, hey, 10% of the time when we try something different, and let's see how that impacts things like NPS and retention. So we look at these 2 different layers. It's a autonomy layer, and we have a whole product group that just focuses on agents that will play the game of marketing and service and really design and deliver customer experiences for businesses. And then we think about the infrastructure needs to run on. And since that's the thing we were founded on, we built -- we started Klaviyo as a database company, that is very good at story indexing consumer data at scale, but making it available in real time or whether it's an agent chatting with a customer or rendering a web page or sending a message. Yes, those are the 2 big opportunities. And we think about that -- that agent layer, I mean, that we think is going to be a bigger business than our current really like software infrastructure business. And then we also think that agents because they're better users, one of the things we talk about is like a typical software user base has this distribution where you like have a couple of power users that are really advanced. And then you have a bunch of intermediate users and you have a long tail of kind of more novelty users. And so you build -- you do a lot of work to try to move people up the curve. But the nice part about agents is they just go straight to advanced mode. In fact, they actually beat our advanced users. They're kind of like PhDs at whatever the discipline in and once we train them up. So we also find that they push the infrastructure and they use more of it. So just as an example, like a typical business for us, we don't have a fee-based model, like we charge based on like the size of the business in terms of the number of profiles or contacts or customers it has. We find that our agents because they're better at delivering customer experiences to those individual consumers, they actually retain more customers. So you end up -- with the infrastructure getting used more. So we look at both of those opportunities. One, we have like a whole business that's going to be even larger than our business today and the acceleration you get from agentic usage. It's most -- I've been since we started Klaviyo because I think both those things are very big opportunities.

Outstanding. So we've expanded into service, right, and expanded the scope, a really competitive environment out there. There's a lot of people who are focusing on the customer service opportunity. There is some really big incumbents, whether it's a Salesforce or the like. And there's a lot of start-ups who see this agentic opportunity as well. How do you look to differentiate the Klaviyo solution from both the incumbents and like the startups coming into the marketplace?

Yes. Well, we're very good builders. So I mean, no surprise, we kind of put our heads together and think like, okay, well, what does it take to build agents and build kind of a good agent building platform? Like I think that's become table stakes, right? Can you -- how do you train up an agent and evaluate it, deploy it, et cetera, across all the various channels. There's 2 things that we're adding to our customer agent that we think are quite differentiated. The first is we built a set of algorithms, you can think of almost like agents on top of the agents that are good at training up customer agents. And let me explain why this matters so much. Like we've got about 193,000 businesses. I'll tell you like there's probably 1,000 that know what it means to build an agent. If you gave them the tools, they would know how to train that thing up. And one of the patterns we've seen is there's a lot of this like, well, we'll deploy engineer or we'll forward deploy engineers or we'll sell services around training these agents up. The problem is like for a lot of businesses, that just immediately price them out of the market. I mean, they can't afford hundreds of thousands of dollars to train this thing up, once, let alone ongoing cost, right? They're just not at that scale. We actually think there's an enormous amount of latent demand amongst SMBs for customer agents because those are also the businesses that are the least optimized when it comes to customer service. Because who is it? It's the founder of the business, the owner of the business or somebody on the team that's handling these conversations one by one. These are not organizations that are like, yes, we've employed a lot of automation, like we've done about BPO. So we need to help them train up these agents. And so what we started was we started this program where we're building agents, you act like a coach or trainer, right? You can imagine like a fitness trainer that like basically get that customer agent for that business into shape. So what does it do? It goes and takes all the information we can find about that business, right, on the Internet -- on the publicly available Internet plus all the connectors that people already have embedded in Klaviyo. It then goes creates basically -- extend the fitness example here for a bit, a workout routine to put it through its paces and figure out how like strong it is, right? How good is it at delivering customer experience, grades it, figures out where it's weak and then tells either goes and fixes those problems on its own or goes back to the user and say, hey, you know what, you're a restaurant, I can't handle all your reservation request because you haven't hooked me into your back end here. So go connect, toast or whatever it is that you're using to manage reservations. And like -- and now I can answer those. So literally, almost we have this like video game style interface where you're training up this customer agent who represents your business, and it just gets smarter and better. Like as you give it more information as it learns on its own. And what's really important about this is we think this is not important -- this matters just for SMBs. It's also really important for enterprise because enterprises need the agents that are going to be sitting on top of their agent kind of like the boxing trainer that's get whipping it in shape as the business evolves, right? A large enterprise might launch a new product line, a new set of services. They might acquire a company. Something changes and they need to like constantly evolve that agent. That's not a job for human developers to do the way we're doing it today. So we're running -- I mean, we're sort of a little bit existential for us. We want to serve all 190,000 businesses. So therefore, we have to build these agents that can train up these customer agents. And we think this technology is going to become table stakes and the differentiator in enterprise as well. So that's one. The second thing is, because we already have our customer database, right, which is -- you think it's like a data warehouse to just optimize not just for analytical queries but also real-time usage, right? You can connect it to it a customer agents, it's not going to go think for 10 minutes. It can respond in milliseconds. Because we already have that, every time you talk to our customer agent, if we have context on you, you can think about, we talk about like memory files when it comes to live ChatGPT or [ Claude ]. We basically have a memory file on that consumer, and we can inject it into the context, into that conversation and optimize it. And just like you see, like if you play with these personal AIs, like, man, it starts to feel like it knows you, we can do the exact same thing. So now you can imagine you're getting these signals from our marketing use cases, what are you browsing on a website, looking at it on a mobile app. How are you using those products and services, interaction with our marketing. We then plumb all of that into the customer agent and it produces a way better experience. Simple example, like we have folks that will come back, come to websites where we've enabled web chat through our customer agent, people just stop browsing the site, they'll just ask for recommendations and get spit out things based on their past purchase history because they're already off through our customers agent, our customer can handle that. It's instantly a better experience. So I think we're going to find personalization on the customer side, not just generic handling, but personalization is going to be another big differentiator.

Got it. And then Chano, in terms of taking this platform and 1 of the things that has always been impressive about Klaviyo is it brings the customer a solution. It doesn't bring them a set of technologies, right? And that's I think a necessity for operating in SMB, SMB customers, they don't care about technology. They don't want to deal with technology. They just want a solution to their problem. Does that resonate in the same way as you go into the enterprise? Or does the go-to-market have to change the presentation of the solution have to change to make this amenable and make it attractive to a larger enterprise customer?

Yes, great question, Keith. What we're seeing with the enterprise customers is they're looking for 1 single unified platform that really powers that entire customer relationship because otherwise, they have too many different flows and architectures and they don't provide that much efficiency and they are costly. Let me give you a couple of examples, right? When they have that kind of retrofitting into generative flightwheel effect and understanding of the market -- of the customers as a whole with that kind of memory file and understanding the consumer that Andrew was talking about and we're helping out provide real-time information with some of our automated flows. Those produce 10x more revenue per customer that basically static customer campaigns will produce, right? So that talks a little bit to the power of the -- clearly, on the infrastructure, right? It resonates, obviously, in the enterprise, I think we commented that we doubled the number of customers to 1 million ARR plus in Q4 last year, and we have our largest enterprise pipeline ever. So really, we are confident that we can solve for that problem.

Excellent. And then, Amanda, to bring you into the conversation, in early 2025, you guys implemented some pricing changes, billing based on total active profiles and auto downgrade feature, flexible sending options. As we go more into the agentic opportunity, are there more pricing changes that need to take place? Or are you comfortable that you will sort of positioning to accrue the value of what edges bring into the equation here?

The beauty of our business model is that we do not price based on seats. We have never priced based on seats. We priced based on the value that we provide to our customers. So our mission has always been that we are going to help you take your most valuable asset, which is your customer relationship and make that even more valuable because we're going to help you with technology, make it feel like every single customer is the only customer that, that business has. And by personalizing those experiences by, as Chano said, making them more automated, we increase the value of those profiles. And by making each messaging and each communication and every experience more valuable, we increase the demand for it. So the pricing going forward will continue to align to that point of view, which is as we help customers generate more revenue and more value, we all align to that. So in the case of marketing agent, which Andrew spoke about, it will be priced based on the number, roughly the usage that you have and how are we able to drive really great outcomes with that usage. So it will be -- think about it as roughly equivalent to tokens. But the most important thing is that, again, if that usage is driving better consumer experiences, customers are very, very willing to pay for it.

Got it. And if we think about that and take it up 1 level into the 2026 revenue guidance, you talked about a minimal contribution from the newest slate of the AI products and the service products. Can you help us better understand the adoption curve, like what should be our expectation for the adoption curve? Klaviyo service was only went into GA back in September 2025. The agents now coming into the marketplace. How should we think about that adoption curve?

The way to think about that adoption curve is off to an incredibly strong start. So service, for instance, is our fastest-growing product in our company history, even going back to looking at text, which has been a really significant product for us over time. And today, almost 30% of our SMB Plus customers are using that text product. Service is off to an even faster start. What's driving customers to adopt it is that as they get exposed to the product, they're seeing the revenue that it drives and they're seeing the benefit and really ingraining it at its workflow. So one of the things we're really excited about on the marketing agent side is many of the customers who are adopting marketing agent, it's become their default. It's become the primary way that they're building campaigns because they see that it drives back to better outcomes. One of our customers, Adelson is seeing 50% higher revenue per campaign from the campaigns that they're generating with marketing agents. And then similarly, on the customer agent side, the customers who are really adopting it into their workflow is becoming the majority of the way that they interact with their consumers. And so as it becomes more embedded and customers are seeing this revenue from them, we're really seeing adoption pick up.

Got it. So we started to already see kind of the benefits of multiproduct adoption. I think you guys talked about 60% of ARR now coming from multiproduct customers, 15% from customers adopting at least 3 products. Any kind of view on where that's going to go over the next 2 to 3 years? And Chano, how could you kind of further accelerate that sort of platformization, if you will, of getting these customers to buy into the broader capabilities of what Klaviyo is able to bring.

Yes. Obviously, as I said before, I believe on the power of that unified platform, and we're going to be setting up things that will be much more ad hoc in terms of really focusing on upfront on those problems for those largest customers in terms of the upselling and cross-selling motions, clearly, thinking much more in advance in terms of pipeline and qualification of those opportunities and what are the ones that we should be participating on. But as Andrew said, the opportunity for some of these products is potentially 5x to 10x what we have today in terms of the current opportunity. So typical customers may start more on e-mail and text, but we're seeing much more adoption moving more rapidly, and we're going to have some cross specialist teams that are going to be supporting some of those motions going forward. And we're closing the loop much better between the value proposition that is coming from product marketing and sales that we were doing before as the teams are getting more ad hoc into how do we drive dynamics that work in the enterprise market.

And a big part of what makes that sale so possible is the value that our customers see from it. Just a couple of examples to help bring that to life, we have 1 brand we work with who's in the D2C space who unified their e-mail, their text messaging and their analytics on Klaviyo and they saw their time to generate campaigns decreased by 60%, and they saw their total cost of ownership decreased by 30%. So they're really seeing the benefit of having everything on one platform. But even more importantly, it's the uplift that it helps to drive from them. Customers who have more than one product with us are seeing much higher revenue coming from it, even higher than just adding 1 plus 1. It's almost a 1 plus 1 equals 3 situation here. And a great example of that is a makeup brand who we work with, who again, unified e-mail, text messaging, analytics and in their case, service on it. In service, they're seeing high 70s resolution rate, and they saw a doubling in their AI-assisted revenue. So we love it when unifying the platform drives higher revenue for our customers.

Got it. So -- and given sort of the -- given the environment that we're in, I mean, software has been under pressure through most of 2025, 2026, it probably even accelerated. And a lot of it is on concerns about Claude work in terms of tool use is about the ability to DIY, the Amazon CTO tweets that he developed the CRM system over the weekend. I'd like to ask him if he wants to support that CRM system every week and for the rest of his life, but that's another question. But all of this has weighed on multiples in software and multiples for Klaviyo as well despite no degradation in terms of the fundamentals, which is frustrating for us, and I'm sure even more frustrating for you guys. But Andrew, where are investors getting it wrong, right? Like why is DIY not a risk? Why should we not [ heed ] more signal from what the Amazon CTO is tweeting out about what you can do with these vibe coding tools?

Yes. Well, look, first of all, we always take a very long-term view of like, hey, what our customers care about, like where is technology really going? So I go back to 2 things, like 2 things we think are like very persistent themes. One, is, hey, we now have the technology that you can build agents that can do the work that humans are doing before. Like that's just a whole new category of software that is going to be won or defined, right? Literally, I mean, if not certainly years like certainly in probably the next couple of months, right? And I think we're going to find out also what -- I think one of the things that's tough is, hey, what software is actually like differentiated infrastructure? And what's like a nice code of paint on top an Excel spreadsheet. And I think we're -- I don't know many -- that's focusing -- you had to kind of figure that out. But like I think that's something that we feel very confident. It's like, oh no, no, what we build in terms of our customer database, what we do around messaging is like, hey, you couldn't just go replace that. So your comments on like -- do yourself, actually an experiment that we run internally to kind of prove this out -- actually, I think in the future, very near future, you're going to get a lot of agents that are actually making the decisions about what software to use. It's not going to be humans doing eval. It's going to be agents -- tell me what you think is the right stack. So 1 thing we do is we actually set up some agents, some coding agents and others. So like, hey, your business, and we do this different sizes, you're just starting out or you're a larger enterprise. And hey, you want to set up like you want to build your software stack so that software defines the entire customer experience. So you're going to want something that is this always on agent, API, voice, text, et cetera, that is our customer agent or something that looks like that. You want something that's going to handle marketing, proactive messaging, something that personalize and customizes mobile applications, the website, even maybe in store experiences. And you want some underlying database that kind of makes all that consumer data accessible, right, and store this, you have a sort of central source of truth, right? We tell have to go out and go build that. And we say, like, and by the way, I want to use as much open source as you can, right? I mean do the best, do high-quality job, but do -- dumb down stack as much as you possibly can. And what we find consistently when we send out [ United Coding ] agents can code for hours and hours. What they go out to do is they basically try to go replicate our data platform and they quickly recognize they're like [ are few ]. Why can't use a data warehouse or data lake, it's actually too slow. I can't use it off-the-shelf database, like it's not flexible enough in terms of the query. Okay, you know what I'll do is I'll build this hybrid stack that basically rebuild what we try to do, right? Now it doesn't have the insights like the query patterns that we've seen, right, and that we have these routing systems that can optimize for that. So I think there's a little bit of an experiment that folks need to run like that of, hey, how really like trivial or nontrivial is that underlying infrastructure. And so we feel really good that what we've done on the database side in terms of real-time access to customer data and it's like attribution loop built in, plus what we've done with messaging around identity and compliance really matters. So on top of that, we're actually -- we're trying -- we're now -- we're giving a lot of work to expose that agent so they could pick us from the start or I think a lot of RFPs now will be done agentically. And then also, we have our agent that we're running that we'll do marketing, right? It's our marketing agent. We'll just go define the marketing strategy for a new business or audit the market strategy of some of our largest customers. And again, it's kind of reasons at a PhD level because we expose it to all the marketing data we have, here is what works best. And then we tell it -- pick whatever underlying technology you want in terms of, like, how would you access customer data, how would you build marketing in these customer experiences, tell us what requirements you need for that system. We're actually using those agents, not only they [ plumbing ] back into obviously our infrastructure, but we're telling us like where are there gaps in what we built that you would want as an agent, if you could have it. And it's actually pressing like hard. So I'll give you an example. Just e-mail at the medium actually allows for a lot of interactivity. We're not used to this because very few companies actually expose it. But you might have bumped into like a Google spreadsheet or Google Doc that allows for commenting in line and things like that. It's actually any business can do that because most people don't have the time to be able to do it. Our marketing agents will happily build interactive texts and they convert way better, right? But don't call back to our system and say, hey, it'd be really nice if you could add support for these different content types and maybe some components like this, so the user can do a last mile editing. And so that's now become part of our road map for our infrastructure. So we're actually using agents almost as like a customer advisory panel to feed back into the road map for our core marketing and now customer service infrastructure teams. So I think -- I would look for companies that are having kind of loops of like they're trying to build both the infrastructure layer and the agent layer and getting them kind of work together to make both parts better.

Got it. So Amanda, it sounds really cool, but it sounds like a lot of tokens, right. And in FY '25, you saw operating margins come under -- gross margins come under some pressure. How should we think about that gross margin line on a go-forward basis? Is this a fundamentally kind of different COGS equation going forward as we get more and more of an agentic layer on top of the infrastructure layer?

Sure. I would think about 2 things when it comes to gross margin. The first is that we have a lot of levers to play with. We have different products in the portfolio as we expand the product portfolio and we're offering new products that enables us to mix out the customer relationship in a way that overall is profitable. And the second is that as we scale, we are getting increasing benefits from our own scale and our own infrastructure, which helps a lot. But if you take a step back on gross margin, the way that we really think about it within our business as we think about almost miniature unit economics for each product line and each 1 that we're offering. So it's not just about the gross margin, it's also about what's the right R&D model to support that. What's the right customer acquisition model to support that. And does each product that we're selling have strong unit economics. So in certain ones, you may have higher -- or sorry, lower gross margin but they also come with lower customer acquisition costs because as Chano said, they're being sold into the existing customer base. And they come with huge expansion potential because they're driving those better outcomes and that higher revenue. And our customers almost see our business as a revenue optimization engine for them. So if we can show them those better outcomes, they continue to expand to the point where they're seeing great ROI. So we think about that gross margin not in isolation, but really for each product, making sure that it's delivering the right economics for the long term.

So then if we abstract back, we see the results of that. Operating margin is up 170 basis points. You're guiding to another 100 basis points of margin expansion in the year ahead. So we're running short on time here. There's a lot more to talk about. So maybe just as a sort of wrap-up question, there's so much innovation taking place at Klaviyo right now. There's expansion of the kind of the market opportunity. As a fundamental analyst, I always want to focus on the fundamentals and the stock price will take care of itself. If we're looking for the key indicators, the key performance indicators that is working, that this is really taking off, what should we be looking for in your results? What should we be looking for in terms of the KPIs to show us that, hey, listen, the fundamentals are following through on the opportunity?

Yes. Well, I'll give you a couple that we look at internally. So -- and we'll start to share more of this as we go. One is we think this agent opportunity is very real. And what we've seen in some disciplines, coding, et cetera, is like is coming to all parts of human labor, but let's say, digital human work. So you mentioned like, hey, multiple products. I mean we actually think about like, yes, multiple products, but like today, it's all multiple products or primarily multiple products through our infrastructure, right? I mean we have a lot of adopters of our customer agent. We actually think it's like by this year, this is the year that every business, small business and enterprise will adopt a marketing agent that helps augment what they're doing, right? The PhD that sits in the room and helps the whole team be better. And then two, this is the year that everybody adopts like a customer agent for their business. So last year was the year of personal AI. This is the year that like every business doesn't just have a website. It actually needs an agent that represents them that's always on. I think both of those are going to grow quite quickly. And so we actually look at like the multiproduct adoption, but really focused on like how many agents of ours have customers adopted. So yes, stay tuned. We've seen some good growth so far, and I think there's -- this is going to be the year that really takes off.

Amazing. Great story going on at Klaviyo. Congratulations on the success, and thank you for coming and sharing with us.

Thank you.

Thank you.

Thank you.