For listeners who might not be as familiar, tell us a little bit about Moveworks as a company. What's the core product and business today? Was it always that way, or can you walk us through the journey of how you got here?

Moveworks is an agentic AI assistant that gets rolled out to the entire workforce, typically inside a large enterprise. We help employees search, find information, get help, and also automate tasks. We do this all within an agentic reasoner, along with hundreds of connectors and plugins that allow us to navigate the enterprise. Ever since folks got inspired by ChatGPT, they really wanted to see that play out inside a corporate setting, and we've been able to deliver that. Of course, we started this well before this hype cycle began, but that's essentially what we do. And we do that at companies like Broadcom, Siemens, and Unilever, large organizations that really want to bring this power to their workforce.

One of the things that came up when I was reading through the Moveworks history was that back when you raised your Series C, which was in June 2021, more than a year before ChatGPT was released, you or one of your co-founders wrote that at Moveworks, you understood AI was critical to the future of work. Why was it so clear that this was the future, and that this was what you were going to lean into with Moveworks?

It was part of our core assumption that AI could be used to really make things simple and make things natural. And conversation as an interface was something we were always fascinated with. I don't know how that began. It's more our own innate intuition that this was such a great interface to bring people of all roles, all departments, and all responsibilities together in a way that could really help lift the work they do and allow them to focus on the more meaningful tasks they need to get done.

So we knew that AI had a central role. How much of that role could be realized only came about as the world also advanced and models, including open-source models, started to evolve. Then we started to see this could actually impact much more than just a certain area of the business: almost every sort of use case that, now with the agentic architecture, we can probably figure out how to get AI to help with.

How do you think about whether the advent of AI helped, was neutral, or hurt Moveworks and the product and company you were building?

The fact that we were not just theorizing or contemplating these concepts. We were actually deploying them to real users. Today we have over 5 million users on this platform hitting it every day, teaching us where it's getting used, where it's not getting used, and what the most powerful use cases are. Even back in the early days, we were smaller, but we might have had a million-plus users even at the start of COVID. You start to track that, and you start to realize the folks here have been seeing that kind of engagement: logs, errors, things where people get stuck. That has really given us a lot of time to harden the system and make it perform in all sorts of use cases.

One of the disadvantages of a brand-new upstart is that they don't have the truth of what actually gets asked, what is actually needed, and what actually drives ROI. With over 350 customers in the eight-year journey, I've been through more procurement renewal cycles than I think any other folks in the AI space. And I can tell you, you can have the best idea and the best product, but when procurement comes around 12 months later and says, "Let's take a look at that impact. Let's take a look at how many seats and licenses we have," you start to get really close to the truth of what you're doing and what you're not doing. That sort of feedback has been instrumental in us figuring this out and getting better and better at it. And I think that's something the folks at ServiceNow saw, as we had a lot of overlapping customers throughout these years.

Perfect jumping-off point. Let's talk about the acquisition. How did the deal with ServiceNow come about? And why ServiceNow? Why were they the company you ultimately chose to partner with?

We found ourselves in a situation where I think over 250 of our customers are using the ServiceNow platform. As I was saying, great companies aren't sold, they're bought. In our case, we've been working in a slightly different realm by focusing on this employee layer, this employee front end, along with the search capabilities. ServiceNow has this incredible platform of automation and agentic capabilities that they've been pursuing with fulfillers, and almost the entire operating system, if you will, for the enterprise. The combination could really deliver even more value to the market and to our customers. We have many who have already realized that value, and we want to be able to bring that to even more.

Jumping off of that, one of the questions Vaibhav suggested I ask you was about the lessons from navigating a company through not one but two black swan events: COVID being one of them, and perhaps ChatGPT being the other. What specific things did you do to position Moveworks to thrive, and ultimately get acquired by ServiceNow, in the face of two pretty unlikely events, one of which enabled what I imagine is a tremendous unlock of competition as well?

COVID was one where all of us were watching to see what it was going to mean for our businesses. We had a very strong point of view that employee experience really mattered, and we tried to go out there and evangelize to people why this is important. COVID actually accelerated that mission, because now if you were even five minutes late to a meeting because of some tech issue, it was frustrating. So companies around the world started to hire people with VP titles in employee experience, chief employee experience officers. And we were like, "This is amazing, because this is exactly the kind of constituent we need on the other end to buy and understand our vision."

So while COVID had uncertainty with it, especially as interest rates started to rise and companies started to clamp down on spend, we were fortunate that the spending was still being directed toward deflationary efforts like automation, which Moveworks played very nicely in. And we interoperate with so many systems that it became very clear to people that we could be a great partner for them.

I'd say the ChatGPT moment, the asteroid that hit the earth, if you will, was more disruptive to, I think, everyone. For us, it was a moment in time where we had this reasoning capability that was more or less single-shot and had best-in-class capabilities relative to the market. But all of a sudden, employees and folks who had their hands on ChatGPT were like, "Well, I want something like this." From a founder standpoint, and you mentioned talking to Vaibhav, we found that we were at a really important juncture at that point. This was probably about two months after ChatGPT came out, and it obviously opened up a lot of new possibilities. So we found ourselves at this intersection around March of 2023, and we had to make a decision, which was twofold. Do we take a look at our core reasoner and rebuild it using this new agentic view of how we think large language models could play a bigger role in planning and replanning actions within the enterprise? Or should we take this new technology, which people were excited about, and apply it toward an area of summarization in search, which became the hype around RAG?

I remember talking with our team and my co-founders, and those two were presented as two paths we could take immediately. The discussion was, which one is harder? And the answer was agentic. And I was like, "So why is that harder?" The truth was, people were like, "Well, there's no guidebook. There's no manual. We're going to have to figure this out. We're going to have to invent our architecture around how to build a truly agentic platform that gives us all these new capabilities that we think the world is going to be looking for." So we went down that path. For the next 12 months or so, the team built an agentic reasoner that was far and away ahead of everyone we had been tracking. Of course, the word agentic gets applied at the language model layer and so forth. We're talking about these large system layers.

In November of 2023, we had our first five customers using the agentic reasoner. Then in February of 2024, we started rolling it out to all of our customers as a new platform they could decide to adopt or not. And that rapidly, in a matter of 18 months, got us close to 5 million new users on this platform, which has really expanded our capability. Of course, we then fast-followed with the deeper version, what we call agentic RAG. We were already doing vectorization of files and such, because we were using these models prior to ChatGPT. But now we could incorporate that into the agentic experience and bring more sophisticated search, summarization, and ranking capabilities to the user experience. So that was our decision. And to do all this, you also have to have the founder's mindset of being able to delete, I don't know, 25% of your code base.

I was going to say, a little bit of creative destruction, right?

Yeah, it is truly one of those bonfire moments where you're like, "Okay, we're going to move away from this approach and go to this other approach." And I think it was that transformation that happened in 2023 that was gut-wrenching. I saw a lot of employees not know what this was going to mean or how to think about it. But we knew that if we could do this, and do it in a hurry, we would be the first to come out with a platform and a system that would be adopted widely by some of the largest companies in the world, which we've been able to do.

There's a really important insider lesson there. This is a tale as old as time. I'm thinking of Zuck and how he handled the transition from desktop to mobile. You burn the boats, and you say, "Look, this is the future. This is the way. I don't care what happened over the past one, two, or three years. We need to do this." And then it's sink or swim. So kudos to you and the team for having the conviction to make that call and move the company quickly in a very specific direction. And look, the end result is not a bad one. Bhavin, thanks so much. It was great to see you, and congrats again.

Thank you, Eric. Thanks for having me on.