How ServiceNow is preparing for an agent-driven enterprise future

We keep hearing how AI agents are going to dominate in the enterprise, but so far the hype has not matched the reality. In part, that’s because implementing AI agents is complex, and many companies are struggling to build them at scale. Enterprise software companies like ServiceNow have been jockeying for position to be a trusted source for implementing this promising new software, competing with SaaS vendors like Salesforce and Workday, large language model makers such as OpenAI and Anthropic and the big 3 cloud vendors.

Amit Zavery, president, chief product officer and chief operating officer at ServiceNow, sees companies trying to build AI agents and agentic frameworks on their own, and he identifies that DIY approach as a point of failure for a lot of organizations. While there is some evidence to support that viewpoint, most companies remain in the AI agent experimentation stage, regardless of whether they are building or buying.

A November McKinsey report, The state of AI: Agents, Innovation and Transformation, found that two-thirds of respondents, a number that has remained persistently steady over time, have not yet begun scaling AI. The report found that a similar number of respondents are experimenting with AI agents.

Although the technology is advancing quickly, there are still big gaps in the infrastructure to make agents truly enterprise-ready. Until those issues are addressed, organizations could continue to struggle to achieve wider implementation.

ServiceNow has tried to tackle this by putting together a more comprehensive solution, partly by building in-house and partly through acquisitions and partnerships.

Building a control layer

One of the reasons companies are struggling, even as the technology becomes more capable, is a lack of cohesive structure to support agent communication, permissions, security, monitoring and other key functionality large companies expect from enterprise software. It's one thing to build a one-off pilot, it's another to have a bevy of agents operating across an organization. So far, few if any companies have achieved that. 

ServiceNow is trying to give customers more control over the process, and has developed a centralized AI management tool called AI Control Tower. Zavery says it takes care of some of the concerns around having an enterprise-ready set of functionality for the agents as they move through the company. 

"What we're doing now, as we add AI Control Tower, is defining for AI agents what abilities they have, what their identity is, what privileges they can take, who are they doing work on behalf of -- all the things we guarantee with a lot of security elements, and security is going to be the core part of this change," he said.

But as Zavery says, AI Control Tower is the scaffolding. They understand they couldn't build everything, especially given the speed at which the market is shifting and changing. That's where acquisitions and partnerships come in.

Strategic acquisitions aplenty

The company realized it couldn’t close every gap by building products internally if it wanted to keep pace with the AI market. As Zavery puts it, the company’s recent buying spree isn't a change in direction, but a boost in speed. "Our 2025 acquisitions don't represent a pivot away from organic growth and building, they represent an acceleration," Zavery said. "We're focused on acquiring strategic capabilities."

That focus resulted in nearly $12 billion in investments across three major deals in 2025. The first came in March with the Moveworks purchase for $2.85 billion—giving ServiceNow an AI‑native capability to transform natural language requests into automated workflows. That record didn’t stand for long. In December, the company agreed to acquire security startup Armis for $7.75 billion, just weeks after buying identity security company Veza for a price reportedly just over $1 billion. 

It also has partnerships with OpenAI, which was announced just this month, and prior agreements with Anthropic and Nvidia to incorporate major large language models directly into the ServiceNow platform and make it easier for customers to work with these popular models.

Zavery sees these pieces as the essential components that sit on top of the AI Control Tower. "What we have with these acquisitions: conversational AI that understands what you need, identity controls that govern who and what can act, security that protects every connected asset, and a control tower that orchestrates all of it," he said.

Even as ServiceNow layers on all of these pieces, Zavery hears the talk like everyone else about how agentic AI could kill SaaS, but he isn't buying it. Perhaps it's self-serving, but he believes his company is well-positioned for the change, and those that aren't will be left behind. "If you are not skilled in AI and don't provide AI-native products long term, irrespective of whether it's SaaS or not SaaS, you will have a problem surviving in the market today," he said.

ServiceNow has clearly positioned itself to compete in this shifting marketplace. While it’s unlikely any vendor will own the full agentic AI stack, the company is using acquisitions and organic products to try to remain as relevant as possible today and well into the future.