How Will AI Agents Redefine the Workplace?

Artificial intelligence (AI) agents are changing how companies can automate work.

The emerging AI agent market is projected to grow from $5.4 billion 2024 to $50.31 billion in 2030, with a compound annual growth rate of 45.1%, according to a report by Grand View Research.

Here, we look at AI agents and how they’re set to disrupt roles across departments and the way enterprises assign and complete tasks.

Understanding AI Agents and Agentic AI

There's a distinction between AI agents and agentic AI, according to Michael Bowers, president of Heed AI Consulting.

AI agents automatically perform specific tasks within a controlled or predefined environment and are typically rules-based with a narrow focus on one job. AI agents “excel at repetitive, well-defined tasks,” such as customer service, automated scheduling and task-based recommendation engines, Bowers says. Unlike agentic AI, AI agents, he says, “don’t autonomously adapt or change their core decision-making framework unless re-programmed by human engineers.”

Because AI agents use AI foundation models, they’re more sophisticated than traditional chatbots and adapt to different scenarios and perform more complex tasks, according to a report by McKinsey & Company.

On the other hand, agentic AI refers to “systems that act independently and exhibit autonomy in their decision-making processes,” Bowers says. Agentic AI systems are built, he says, to “perceive their environment, analyze data, make decisions and adapt over time based on the outcomes of those decisions.”

Agentic AI systems are more complex and designed to solve problems in unpredictable or dynamic environments and evolve with experience, such as self-driving cars, automated trading systems and AI-based medical diagnostics, Bowers says.

AI Agents Taking Over Tasks

Major software companies, such as Microsoft, Salesforce and ServiceNow, are rolling out AI agents as the next phase of AI innovation.

Magentic-One is a generalist multi-agent system by Microsoft that’s designed to solve web and file-based tasks in the digital workplace, including SharePoint agents, facilitator and interpreter agents for Teams, project manager agents and employee self-service agents for HR and IT.

Agentforce is the “digital labor” platform by Salesforce that allows companies to create specialized AI agents to support employees or customers and execute tasks, such as for customer support, e-commerce personalization, marketing campaigns and sales outreach.

In the Now Platform by ServiceNow, the company integrated AI agents for tasks in procurement, software development, HR, customer service and IT service management, “keeping people in the loop” for oversight and governance.

AI agents can be “a wonderful addition” to business workflows, said Myles McPherson, an interaction designer at Influence Labs.

“I build them a lot at my agency, and they play one crucial role: automating tedious work that would take a person hours to do, like filtering, moderation, review and approval and so on,” McPherson said.

“When built correctly, AI agents can be like little virtual assistants.”

Related Article: Can Agentic AI Revolutionize CX and EX?

Developing and Managing AI Agents

While AI analysts pretty much agree that AI agents will be AI’s next big thing, they differ in their predictions. Deloitte predicts that next year, 25% of companies that use generative AI will launch agentic AI pilots or proofs of concept.

Software companies rolling out AI agents expect the automation technology to advance.

“Over the next year, we will witness the evolution of enterprise AI agents as they become increasingly sophisticated in their reasoning and comprehension capabilities,” said Dorit Zilbershot, VP of AI and innovation at ServiceNow.

In particular, Zilbershot expects to see AI orchestrators that serve as the link between agents and humans. Like the conductor of an orchestra, she said, an AI orchestrator coordinates and manages communication and interaction between agents.

AI orchestrators allow AI agents to share information and hand-off tasks efficiently, which is essential for managing complex workflows — and makes it easier for humans to manage teams of AI agents, Zilbershot said.

Zilbershot believes AI agents will become fully autonomous by the end of next year.

“Expect to see AI agents independently automating complex, multi-step processes without a human in the loop,” Zilbershot said.

But Forrester is more pessimistic about AI agents.

“They’re not ready yet,” says Forrester in its "Predictions" report on technology and security. Forrester expects it to be another two years before AI agents could meet user expectations because of their current complexity.

In fact, Forrester predicts that 75% of enterprises that attempt to build agents themselves next year will fail. Instead, the firm says, those enterprises will turn to consultancies to build custom agents or use agents embedded in their existing vendor software ecosystems.

Co-Existing With AI Agents

There are fears among many workers that that AI agents will take their jobs as the tech assumes more tasks previously done by them.

For instance, in an assessment of 22,000 tasks in the U.K. economy, 11% of them are exposed to generative AI, such as the large language models (LLMs) that underly AI agents, according to a report by the Insitute for Public Policy Research. The percentage could increase fivefold if AI systems, such as agents, beocme more deeply integrated into organizational processes.

However, desk workers spend 41% of their time on tasks that are low value, repetitive or lack meaningful contribution to their core job functions, according to a report by Slack. AI automation could help workers refocus on higher-value activities.

“The tension between human and machine is timeless,” said Sawyer Middeleer, chief of staff at Aomni.

“There was similar angst when electric typewriters and later, computers entered the workplace. In all of these instances of technological disruption, it’s both true that economic productivity grew and also jobs were lost.”

In response to employee concerns over AI agents, a common theme is that AI agents won’t take your job, but people who know how to take advantage of AI agents may — if you don’t learn to use the technology yourself. That said, it’s important for companies to acknowledge and assuage employee concerns.

“The least leaders can do now is recognize the understandable anxiety of workers who see their jobs becoming redundant to new AI software,” Middeleer said.

“There’s something different about AI compared to other technologies. Because these systems can write, speak and reason in human-like ways, many people make the mistake of treating them like people.”

For instance, he said, marketers and salespeople are describing AI agents as AI employees.

In fact, the human resources software company Lattice received pushback this summer when it said it was going to treat AI systems like employees.

“Within Lattice's people platform, ‘digital workers’ will be securely onboarded, trained and assigned goals, performance metrics, appropriate systems access and even an accountable manager,” says Sarah Franklin, CEO of Lattice, in a LinkedIn post.

The resulting outcry caused Lattice to back off.

“It sends a message to your employees that you value them the same way you value your software — as line items on a profit-and-loss statement — and that you don’t care about the intangible humanness they bring to your team,” Middeleer said.

“It shatters the veil of workplace values and culture that most leaders at least claim to support.”

Related Article: Will Your Next Hire Be an AI Agent?