A New Layer To ServiceNow Agentic AI: Exec Gaurav Rewari On The Data.world Acquisition
“We said, ‘OK, what can we do to help our customers and partner with them on their journey to getting their data estate AI-ready?’” says Gaurav Rewari, senior vice president and general manager for data and analytics products at ServiceNow.
AI-focused digital transformation technology developer ServiceNow, which Wednesday unveiled the planned acquisition of data.world, is looking for that company to add a big missing piece to its agentic AI capabilities.
Gaurav Rewari, senior vice president and general manager for data and analytics products at Santa Clarita, Calif.-based ServiceNow, told CRN in an exclusive meeting during the ServiceNow Knowledge 2025 conference that his company realized there were three things it needed to help its customers become AI-ready.
ServiceNow already offered two of those: its AI-ready RaptorDB to make data AI-ready and its Workflow Data Fabric for bringing data from across the enterprise regardless of the source into ServiceNow.
[Related: ServiceNow CEO: ‘The Destruction Of Time Is Like The Ultimate Enemy Of Humanity’]
The third, Rewari said, is making sure that data used by ServiceNow to automate customer workflows via agentic AI is completely trusted and follows a business’ governance requirements.
“It’s not just about the quantity of data that you allow AI agents to access,” he said. “It’s also about data quality. It’s about data trust. Can you trust that data? Do you understand its lineage? Has it been properly governed? Do you understand the relationships between those data feeds? And that third piece of the puzzle, this whole notion of data governance and catalog, that’s the piece we didn’t have.”
Rewari also discussed how the data.world technology will help ServiceNow make a bigger play for AI-based workflow automation along with the security implications of bringing a new technology into the ServiceNow platform.
There’s a lot going on at ServiceNow and its push towards workflow automation. For more, read the entire conversation with Rewari, which has been lightly edited for clarity.
Why did ServiceNow acquire data.world?
Maybe I can sort of paint a bigger picture about how we’re thinking about data and analytics in general, and from there progress to the rationale behind the acquisition. As we thought about what we’re doing with agentic AI, Bill [McDermott, ServiceNow CEO] has been talking about how the other disruptions that we’ve seen with mobile and cloud almost pale in comparison to what we’re seeing now, not just in terms of the scale, but also in terms of the velocity at which companies are moving. And so we thought about it. There’s a line I used in my keynote about, the road to agentic AI heaven often goes through a data hell, and it’s borne out by statistics from Gartner. There’s a focus group that Gartner ran not too long ago that said 4 percent, only 4 percent, of technology leaders felt their data was AI-ready, and that 60 percent of AI projects will fail by 2026 because the data is not AI-ready. We said, ‘OK, what can we do to help our customers and partner with them on their journey to getting their data estate AI-ready?’
What did you find?
We identified three things. The first is the number of complex transactions and workflows that will have to be processed because now there’s machines also doing workflows, not just humans, is going to increase by one or two orders. And these are thinking machines, so we need to be able to support analytical processing in addition to operational processing. We were very early in AI. We were exposed to what that future world might look like much earlier than many other companies. So we invested in our next-generation AI-ready database called RaptorDB. It’s been available for about three quarters, and we’re continuing to invest in it, grow it.
Number two is, you know your AI agents are only going to be as smart as the data you expose them to. And if it’s just ServiceNow data, then we’re missing a trick. Customers are not going to get signals from the other data sitting enterprise-wide. And that’s why we built Workflow Data Fabric. With Workflow Data Fabric, we gave them access not just to the data in ServiceNow, but also to external data sources, enterprise-wide data, data that’s still inside the enterprise. And this could be data that you bring in and put in RaptorDB, or you can let it stay in place. So we want to throw open the surface area beyond just the data that’s in ServiceNow. And we can either move that external data sources into RaptorDB, or we can let it stay in place in a database, in Snowflake, in Google BigQuery, in Amazon Redshift, and the technology that makes that possible is zero copy connectors. Wednesday, we really expanded the frontier of integration in all flavors, either the connectors to where the data comes across or where they stay in place. We are calling that the Workflow Data Network. And it’s like the arms and legs of the Workflow Data Fabric. It fans out and allows you to tap into data sources across the enterprise. Why? Because you need to have your AI agents learn what they need to learn so they can do what you want them to do.
And the third?
Number three is, it’s not just about the quantity of data that you allow AI agents to access. It’s also about data quality. It’s about data trust. Can you trust that data? Do you understand its lineage? Has it been properly governed? Do you understand the relationships between those data feeds? And that third piece of the puzzle, this whole notion of data governance and catalog, that’s the piece we didn’t have. And that’s where data.world comes in, because it’s a knowledge graph-powered data catalog and data governance solution. They have a large number of very sizable customers, so they go after large enterprises as well. And so we decided they would be a perfect fit, because they give us data cataloging and data governance capabilities so you can trust the quality of the data. It’s architecturally built on a knowledge graph, which is a core architectural principle of ServiceNow pretty much from day one. That’s why we made the move. With these three things, we make the data estate AI-ready, so you know your path to AI agentic heaven goes through not a full-blown data hell, but is less painful.
What else does ServiceNow need to make that path even less painful?
CEO Bill McDermott likes to talk about ServiceNow as the control tower for the enterprise. From the first time I heard that, as I’ve had more conversations with customers and understood our products more, there’s a lot there to be unpacked. This week we announced the AI Control Tower. In a similar way, on the data side, we want to think about a data control tower which will have a data catalog and data governance. But there’s some other pieces that over time would make sense for us to add, either through organic development or through via partner relationships. But philosophically, I think that our approach is that we will make sure if a customer has some of these other pillars already in place, that they can be plugged in.
As an example, things like data quality, observability, etc., we want to be sure that we allow those capabilities to be plugged in, or we may choose to provide them ourselves as well. That’s how we’re thinking about it. Our vision is to be a data control tower for the enterprise in its journey to make its data estate AI-ready.
How much work is involved on ServiceNow’s side to make data.world’s data governance, data catalog, and so on technology part of ServiceNow?
The acquisition hasn’t closed yet. We expect it to close sometime in the summer. It just depends on how long the process takes. But it’ll be sometime mid-summer, most likely. We are a one-platform company, so we’re going to integrate it into the ServiceNow AI Platform. That’s the new branding for the platform. And we don’t have a time estimate for that yet, but we expect we should be able to do it very quickly. There are lots of pieces that we want to integrate, like the metadata collectors, the population of the knowledge graph, the core data cataloging and data governance pieces. There’s quite a bit there, but we think we can do that on an accelerated timeline. Unfortunately, I can’t tell you what the timing is until we close, because we take usually three weeks to quickly spec that out. But we’ll be running very hard to get it in the hands of our customers as soon as possible, and it will sit within Workflow Data Fabric. Suddenly, Workflow Data Fabric is about integrating the data and then putting this layer of meaning on top of that. So it’s connect, understand, and then you can take action using the rest of the ServiceNow platform.
ServiceNow is focused on helping businesses automate their processes with AI and agentic AI. When you bring in a new acquisition like this, there has to be a certain level of trust that the technology that you’re bringing in meets the same trust requirements that customers have with ServiceNow. How do you make sure of that? Yes, data.world already has marquee customers. But how do ensure trust as part of ServiceNow?
Look, we have been talking to them for some time, and we have spoken with a bunch of other vendors in this space. And we came away very impressed with the fact that they are built on a knowledge graph. We’re very impressed with the types of customers they have, marquee names and larger enterprises. We came away extraordinarily impressed with the team. It’s like a labor of love for this area. They have, in a good way, an insane love for this whole data, governance, data, cataloging area and finding relationships in the data and capturing that through knowledge graph. We felt that at the outset of our conversations. We talk about how ServiceNow has kept its original startup kind of play. So in many ways, they are cut from that cloth, so they will really fit in terms of the quality of their product.
In addition to the initial diligence between the signing of the term sheet and the announcement, we did much deeper technical diligence to basically get comfortable, and everything checked out. I will just add one more thought, which is that one of the things that was very striking to us about data.world was that, at least from our research, it appeared that they have more users per customer than anyone else in this field. So customers are voting with their feet, and that tells me something about, quote, unquote, the ‘standard’ of the product or the ‘quality’ of the product. It’s a second order signal, but it’s still a powerful one.
