Three professionals set to thrive in the AI agent era

The rise of autonomous business is creating new roles for professionals who can build, supervise, and direct AI agents. According to technology leaders interviewed by ZDNET, three groups will be especially valuable: tech-savvy enablers, agentic course correctors, and outcome-focused collaborators.

AI agents are increasingly being used for tasks ranging from basic operations to decision-making, including the ability to discover, negotiate, and transact. That shift has prompted concern about reduced demand for professional and IT roles. The executives cited by ZDNET instead expect workplaces to combine human expertise with agentic systems.

Tech-savvy enablers

Alex Read, senior enterprise product manager for data at energy provider EDF UK, said strong internal technical capabilities remain essential. His data team supports more than 1,000 users and operates at the center of a federated hub-and-spoke model: data and digital specialists focus on technical enablement, while business units focus on creating value.

“Context is king for AI. The role that we see our engineers performing with agentic and AI-assisted experiences is that they need a high skill base to build the surrounding capabilities that interact with these services.”

Read’s team uses Snowflake Semantic Tables and Horizon Data Catalogs to define data assets in ways AI agents can understand. It then builds capabilities with Snowflake’s CoCo coding agent on top of the data platform. One tool helps service-center staff manage customer queries by pulling information from Snowflake and presenting insights inside Slack.

Engineers still refine the output. Their job may involve enhancing what CoCo produces or tailoring it to make the result more accurate, Read said.

Recommended reading

AI workloads are forcing a transport-layer rethink

At Capita, chief AI and product officer Sameer Vuyyuru is developing an AI Catalyst Stack focused on process observability and measuring how workflow automation can improve operational efficiency. He urged professionals to strengthen their technical skills and start using AI regularly.

“Start working with AI every day. When we look at our thousands of employees, we’ve made a significant investment in giving every one of them access to AI tools.”

Agentic course correctors

Autonomous systems cannot operate without domain expertise and oversight. Read said EDF’s IT professionals work with subject-matter experts who may have 20 or 30 years of energy industry knowledge. Those specialists help agents become more productive and accurate, while technically skilled professionals identify and correct errors.

Stephen Wood, chief operating officer at Rathbones Asset Management, described the change as a move toward relationships in which people oversee their “agentic colleagues.” Humans remain responsible for final checks in a highly governed, process-oriented business.

“What we really want is a world where we’ve got experts who know their stuff, they’re the humans in the loop, and they are the final oversight, the check-off.”

Wood expects friction as organizations adapt, but said companies and professionals that embrace agentic AI and learn to advance its use will be best positioned to survive.

Outcome-focused collaborators

Dan Cherowbrier, CTO at Formula E, said professionals must begin with the business outcome rather than simply automate existing human processes, departments, and silos. In software development, that means applying engineering expertise earlier to define requirements and guide agents, rather than relying on human expertise primarily during coding and testing.

“If you can’t understand what you need to do from the outset, then I think you will struggle.”

Freshworks CTO Murali Swaminathan similarly advised companies to prove that an agent delivers a repeatable result before deploying it broadly: test a few use cases, build confidence and trust, and then roll the service out to everyone.

Louise Newbury-Smith, head of UK&I at Zoom, said the process must remain human-centered. Her view is that AI should empower people rather than replace them—a distinction that will determine who remains effective as agents become workplace colleagues.