When asked what IT job will look most different in three years, tech leaders said the following:
Software engineering, IT support, and identity and access management roles are all being reshaped as AI agents must be provisioned, secured and governed.
IT work is moving from hands-on execution to orchestrating AI systems, managing agent workflows and directing outcomes.
New priority skills are emerging around AI governance, security, cost efficiency and enterprise orchestration.
IT roles are becoming more strategic and judgment-driven, with increasing emphasis on decision architecture, operating model design and knowing when and how to apply AI.
AI may not eliminate as many IT jobs as headlines suggest, but it is poised to alter many of them.
As enterprises move beyond experimenting with AI and begin embedding it into everyday operations, IT leaders must prepare employees for jobs that increasingly revolve around supervising, governing and orchestrating it. The shift has implications for hiring, upskilling, organizational design and even which IT roles become more strategic over the next few years.
To find out which IT job will change the most, TechTarget asked six senior IT leaders the same question: What IT job will look most different in three years because of AI? No single role emerged as the consensus answer. Respondents cited software engineers, infrastructure engineers, IT support professionals, and identity and access management administrators. Yet they largely agreed that the future of IT work is less about execution and more about judgment, governance and orchestration.
What IT job will look most different in three years because of AI?
"The biggest shift will not be job reduction but around how IT engineers work. Most of their collaboration will happen with agents rather than people. Engineers will manage small teams of agentic colleagues, while team-level human-to-human collaboration moves up a level, with people staying in the loop for oversight, judgment and coordination rather than execution.
Engineers who know how to reduce token consumption … will be sought after and paid well.
Harish VundavalliSenior technical architect, Strategic Education, Inc.
The demand for skills will shift rapidly, which we are already seeing. I expect roles to cluster around areas like security in an agentic environment -- more about how you secure and govern agents across the enterprise, and how to do so cost-efficiently. Engineers who know how to reduce token consumption and design workflows that use AI economically will be sought after and paid well.
But the biggest challenge I see is orchestration decisions at the enterprise level. How do teams benchmark and decide which workflows or processes actually need AI orchestration, and which do not? Getting that judgment right is hard, and that is exactly where many IT engineers will reskill and reposition themselves over the next few years."
-- Harish Vundavalli, senior technical architect at Strategic Education, Inc., a higher education services company
"The role that changes most is the hands-on infrastructure engineer -- the person who today writes a firewall rule by hand, configures the switch and builds connectivity one line at a time. In three years, that's not the job. The barrier to that kind of expertise is being leveled. You can already tell a Gemini- or Claude-style interface, 'I need a firewall rule that does the following,' and it will crank out the recommended action and, increasingly, write the rule for you. Push it up a level by telling it, 'Build me an app that lets me control all of these firewalls easily,' and the large language model does that, too. So, the value of the role shifts from 'Can you write it?' to 'Can you orchestrate the systems that write it?'
Some of my strongest IT colleagues are now among the heaviest users of our engineering AI tools.
Joe Wilson CIO, CSG
It's like going from line cook to executive chef. The line cook chops every ingredient and makes each line of code by hand. The executive chef orchestrates with multiple agents producing multiple components, and puts the finishing touches on, making sure it all holds together. We're already seeing it. Some of my strongest IT colleagues are now among the heaviest users of our engineering AI tools, because once you can solve a problem one time and have it apply many times, you've moved from writer to editor.
What decides who thrives in that shift is curiosity. If you're doing something time-consuming every day and you're not asking, 'I wonder if AI can solve this,' you're already behind. The skills, the workflows, even the titles will follow, but at the individual level, the differentiator is whether someone is wired to keep asking that question."
-- Joe Wilson, CIO at CSG, a customer engagement and revenue management software company
"The job that will look most different is the IT manager responsible for enterprise platforms and digital transformation.
The role won't disappear, but its center of gravity will move. Today, many IT managers are still measured by system delivery, vendor coordination, support performance, budget control and governance. In three years, the best IT managers will be judged by their ability to orchestrate people, platforms and AI-enabled workflows so the organization can make better decisions faster.
The IT manager will become less of a traditional service owner and more of a decision architect. Technical knowledge will still matter, but the differentiator will be operating model design: data ownership, process architecture, decision rights, cybersecurity, traceability, AI governance and adoption. The IT manager who only manages tools will lose relevance. The one who can connect AI to business outcomes while controlling risk will become much more strategic."
-- Sergei Irisov, head of IT & digital transformation at ZeroAvia, a hydrogen-electric aviation company
"The software engineering role will look most different because AI is changing the job from producing code to directing outcomes. The best engineers will not simply be the fastest coders -- they will be the people who can decide what needs to be built, break the work into the right pieces, direct coding agents effectively and know what is production-ready versus what only looks convincing.
The best engineers will not simply be the fastest coders -- they will be the people who can decide what needs to be built.
Ken RingdahlCTO, Emburse
Product engineering will become more central to the role. The concept is not new, but AI will make it more prevalent and more valuable. As AI takes on more of the mechanical work of generating code, tests, documentation and prototypes, engineers will need to spend more time connecting customer problems, business context and technical execution. The most valuable engineers will be those who can make strong product and technical tradeoffs, not just complete predefined tickets.
Engineers will also need to become skilled orchestrators of AI. They will still need deep technical understanding to write, review and operate production-grade software, but their velocity will increasingly depend on how well they can guide coding agents through design, implementation, testing, debugging and refactoring. Just as important, engineers will need to know when to use AI, which AI to use and how much AI to use. As AI costs rise across the software development lifecycle, using AI efficiently will become part of the engineering craft.
-- Ken Ringdahl, CTO of Emburse, a travel and expense management software company
"I see my colleagues' roles moving from production-oriented to orchestration-oriented. They no longer code but manage and instruct agents every day. People with focused, narrow skills will need to become broader knowledge operators, knowing the essentials across many more IT areas, so they can work efficiently with AI agents.
Twenty years ago, IT roles had a broad spectrum of skills -- from networking and system administration to backend and frontend combined. Later, the field became more narrowly specialized, but now the tables have turned again, with a need for broader skillsets to manage teams of AI agents. IT teams will need more business skills rather than strict technical knowledge so they can understand broader business and operational needs and translate them into AI agent instructions.
IT used to be the mechanic in the basement. AI is turning IT into the air-traffic control tower.
Boris KolevGlobal head of technology, JA Worldwide
One of the roles that will look most different is IT support -- even the name may start to change. 'Support' sounds reactive, as if the job begins only after something breaks. With AI, this function will become more proactive, evolving into something like a 'digital experience orchestrator' or 'AI service concierge' -- someone who does not simply close tickets, but designs how people get help, maintains knowledge and ensures AI safely handles the first layer of support.
IT used to be the mechanic in the basement. AI is turning IT into the air-traffic control tower. The systems are moving faster, more autonomously and across more routes. The value of the human role lies in ensuring that the entire digital airspace is coordinated, safe and understandable."
-- Boris Kolev, global head of technology at JA Worldwide, a non-profit organization serving entrepreneurship, financial literacy and work readiness education
"The role that changes the most is the one nobody puts on a conference slide -- the identity and access management administrator. Today, that job is mostly human plumbing. You provision and deprovision employees, run access reviews and spend a lot of time working out who has access to what and why. Over the next three years, the human side becomes the easy part. The hard part is AI agents, which are just non-human identities with credentials, except they request access on their own, act at machine speed and can spin up more agents to do the same.
I spent years at Netflix detecting compromised identities in the cloud and, more recently, at HashiCorp, securing the infrastructure and machine credentials behind modern platforms. An over-permissioned agent looks a lot like the thing I used to hunt. So, the day-to-day activities shift from onboarding people to governing software that behaves like people: what an agent is allowed to touch, for how long, under what conditions and how fast you can pull its access when it does something you did not expect.
From where I sit, that makes these roles more important, not less. An over-permissioned agent has a much larger blast radius than an over-permissioned new hire, and it will not trip the human tells we have relied on for years."
-- Will Bengtson, CISO at C1, an identity governance software company
Tim Murphy is a site editor and writer for the IT Strategy team at TechTarget.
