Agentic AI is set to change how business gets done

Enterprise information technology is regularly upended by emerging technologies that force a rethink of how work gets done. Past examples are client server, cloud, and mobile. Today, agentic artificial intelligence is causing a shift in how technology teams—and the business units they support—operate. Tech leaders at ServiceNow expect this change to be more rapid and have a greater impact than any shift that’s come before.

“AI will bring a new level of intelligence, which will help employees do a lot more work,” says Michael Park, senior vice president and global head of AI go-to-market at ServiceNow. “It’s going to supercharge employees to drive a lot more productivity.”¹

One especially powerful use case he’s seeing is new-employee onboarding. Every new hire needs access to human resources systems as well as tools and data specific to their role. In the past, new workers would engage with a range of helpdesk workers, retrieve passwords, log into different systems, and assemble the credentials they needed to start doing their job. Now, AI enables the HR systems to learn more quickly what new hires need and automatically provision access by the respective employee’s start date.

This automated learning approach can be applied to all sorts of business processes, such as case summarization, meeting-note generation, and knowledge-base creation, Park says. Automating these tasks through generative AI capabilities such as summarization, notes generation, conversational chat, AI search, and task automation may save two minutes or two days, depending on the use case. Once they offload simple workloads and tasks to AI bots, enterprises can redeploy workers to more valuable tasks, like improving service levels, driving margin growth, or developing new product offerings—a trend ServiceNow is seeing with its customers. 

“AI in core systems is merely a new capability, a tool to be employed,” Park says. “The bigger strategic imperative is using these new capabilities to redefine the status quo for exponential value creation versus just bringing over existing processes onto a new technology capability.”²

In spite of the rapid progress made in the field of AI over the last couple years, we’re still likely at the early stages of what will be a true leap in how work gets done. The development of large language models (LLMs) has enabled enterprises to build agentic AI tools, which can autonomously make decisions and take action to accomplish user-defined objectives. Chris Bedi, chief customer officer at ServiceNow, thinks this application of generative AI will lead to substantial efficiency gains for businesses.

As Bedi says, “Agentic AI cannot completely take the place of a human, but what it can do is work alongside your teams—handling repetitive tasks, seeking out information and resources, doing work in the background—24/7, 365 days a year.”³

In this agentic future, Bedi sees teams of agents working together with each model, owning its own area of expertise. For example, one agent may manage workloads in cloud environments. Another may handle service-order provisioning. When a job needs more cloud resources, the first determines the requirements and clears it with the second.

These domain-specific agents could be an upgrade over existing robotic process automation tools, which do some of the same things but are also brittle and can’t adapt to changing environments. Domain-specific agents will be better at proactive execution, for example, seeing that a business has a new employee onboarding next week and automatically provisioning the employee’s access to facilities, entering them into payroll systems, scheduling their training, and executing across all the necessary areas.

“That’s where things are heading,” Bedi says. “We’re already seeing glimpses of it. Agentic [AI] answers the question of how we actually get work done.”⁴

However, enterprises aren’t going to get to this agentic future without putting in some work. Park says most organizations will need to invest in data management and governance because maintaining sufficiently large, clean, and organized data sets is the key to getting the most out of LLMs. Many businesses have too many data silos, and highly customized code environments make it hard for AI tools to reach across databases.

“LLMs operate efficiently against large data sets,” Park says. “If you don’t have advanced data sets to drive inference, it’s garbage in, garbage out.”⁵

Park feels this work will be worth it, though. He believes agentic AI will unleash significant gains in productivity, which will change the way businesses connect with their customers and drive revenue. AI is the tool, but the purpose is growing the business.

“It’s about building an intelligent platform that helps people transform business, not just doing AI for the sake of AI,” he says.