Klaviyo, Inc. Citi’s 2026 Global TMT Conference

Yeah, totally. We started Klaviyo in 2012. Before Klaviyo, we were working for a startup building large-scale business intelligence software. So, think of folks like Bank of America, Walmart, Starbucks. One of the key problems that my Co-Founder, Andrew, and I were focused on there was how do we unify all the data sets that exist across these large orgs? If you think about a large coffee chain, the system that knew what the sales were on a given day was totally separate than the system that knew what inventory in the store was certainly different than the system that knew who actually was working in the store that day, and it was different than the real estate system that we could use to look up whether or not it was a college store or whether or not it was raining near that store.

That was the problem we had been focused on, and we had this belief that, hey, businesses in the future will need a single, consistent data store that unifies all this different data and can be accessible at scale and with very low latency. We went through a few different ideas, but built that data layer and then went saying, okay, we knew we wanted to bootstrap the business. Where can we use this system to drive revenue? Eventually focused on the SMB, because we could have very tight feedback cycles, and focused on revenue generation. We early on said, "Let's take this data layer and attach marketing to it." At the time, specifically email.

What we did is of our first 10 customers, a variety of SMBs, a parking lot and software companies and e-commerce, 3 of the 10 were e-commerce companies. We said, "Okay, if we are going to bootstrap, we have to lean into these." Leaned in heavily and said, "Okay, we are going to be the revenue generation platform, the marketing automation platform behind these," but the difference is that we started with this data layer, that is really where our expertise is, and bolted this data layer atop. Fast-forward 12 years, it is really the same. At the time, we were always agnostic to what the channel was. We picked email because that was primarily what our initial customer was using to drive revenue. Over time, it is we think about SMS the same way, WhatsApp the same way.

Honestly, APIs, the website, really anything that can be used off this common data layer to power the business forward, but specifically the revenue. In terms of myself and my role, have done kind of everything over the course of it, from initially co-CEO to Chief Product Officer to Chief Marketing Officer, and then now I spend most of my time kind of across the business, but mostly on product and AI and things like that as the Chief Strategy Officer.

Yeah, it's wild. I think we've always thought of software as you build software to drive outcomes. When we thought about in the early days, it's like, okay, we could spend some amount of time building software ourselves and you always want to have something where you can measure the impact of what you're doing. For us, that was always the revenue of our customers. We still use this as a metric we call KAV, so Klaviyo Attributed Value. Over time, obviously, the company got a lot bigger, many more engineers, and we used to much more think, "Okay, hey, there's efficiencies we can drive with how we let people work, so we can build common engineering infrastructure. We can train people, mentor and guide them.

We can control how senior people we hire are." I think what we've seen over the last few years is dramatic improvements to just how fast we can ship things, and so it's really speed. We've seen 2X changes in how fast we're shipping things off common things that match consistently. We really think there's just radical more. This will just keep improving over time. It's not just in engineering, it's related, it's product design, but it's also across the org. We've seen internal tools we've built and just complete changes in processes that we can almost entirely eliminate internally using different AI tools and every few months, the pace just quickens.

Yeah, review and documentation, but also things like, hey, how do we process expense reports internally? How do we process certain types of customer requests? Where can we automate certain types of support queries that come in? But then all the way down to just literally how do we design product and which pieces are humans doing? There's things where we can now spend more human time figuring out the customer needs and exactly what we're trying to build and less human time on the actual building of or writing of documents or doing things like that, where we can actually shortcut a lot of steps that are now done primarily by AI.

Yeah, totally. We started, we built this data layer, and then we layered the marketing on top. So we always thought, hey, there's a natural expansion that will be channels. So, what are the channels people are using? We were heavy on email. We've since been much heavier on the full suite of kind of outbound, think about it as like SMS and WhatsApp and things like that. More and more on the just what are all the surfaces people interact in. So, not just those, the messaging, but also the website. We've always thought of this as like, hey, we started with the SMB, and for a long time we've been stretching that up to much larger and larger enterprises. So there's a piece of the roadmap, which is just that continuance of the vision.

A second is a couple of years ago started to say, "Hey, really you want the common data and contextual layer, not just beneath marketing, but also against kind of any customer interaction." This led to the help desk and Customer Agent and kind of any products that help answer customer queries as they come in or have those dialogues, active dialogues with customers. That's kind of the second bucket. I think that the third though is, we've always thought of Klaviyo as like the reason people use Klaviyo is to, again, to kind of drive a customer outcome. What's possible now with AI tools is some of that usage is going to be a human, but a lot of that usage are also going to be agents. Some of that happens internally, some of that happens externally.

These are things like, "Hey, look at my strategy and figure out what are the gaps where I could be generating more revenue. Look at my strategy and figure out if I've done anything wrong, where I've made a mistake and can set things up differently. Hey, my designers built this copy. Can you tweak the content to make it more effective? Can you design new content?" Just really the full suite of how do we have an agent sit side by side with a human and make the results coming out of Klaviyo more effective. The key iteration of this is our Composer product, which we released a couple of months ago. We've already had 138,000 customers use it, so kind of tremendous adoption.