CIO Report Series: From IT Admin to IT Engineer

CIO Report 2026 series: insights from exclusive report, based on interviews with IT leaders at Cloudflare, Cursor, Lyft, GitLab, Nextdoor, Zip, Rubrik, IMC Logistics, TrendAI, Palo Alto Networks, and Zscaler.

Most IT leaders know AI is going to change their team's job. Fewer have decided what the job becomes.

The org chart won't answer that question. The ticket queue never did either.

The job the queue defined

Forty to sixty percent of the IT workday goes to ticket management. Most IT functions were built to absorb that load, help desk staffed for volume, escalation paths designed for the overflow, leadership managing the throughput. That structure worked because human hours were the resource being allocated. The bottleneck was always the queue, so the org grew to staff the queue.

Manu Narayan, CIO at GitLab, joined with a clear read on what most IT teams were actually doing with their hours. Volume management. Access requests. The kind of work that kept the team busy and the business largely unaware of what IT was capable of producing.

"AI just has this really transformational ability to say, let's take that mundane work out," Narayan says. "AI can really master that and handle that while our team members can focus on their core responsibilities."

The phrase "core responsibilities" works hard. For most IT teams, those responsibilities were defined by the queue, because the queue was always there, always full, and always winning the argument about where hours should go.

When AI removes that volume, what's left is the question the queue had been putting off for years.

Where freed hours go by default

The default answer to freed capacity is more reactive demand. It just arrives in a different form.

Shyam Bhojwani, CIO and CISO at Nextdoor, ran IT and cybersecurity simultaneously for 15 years. When AI became a visible business priority, his team became the destination for a new category of inbound requests: AI tool evaluations, enablement questions, implementation support, compliance reviews. All legitimate work. All capable of consuming the same hours the old queue consumed.

"We were getting a lot of requests," Bhojwani says. "Our focus completely shifted from what we were doing to, what is this new shiny object? How do we tackle this?"

A new bucket opened alongside the existing ones. The operational model expanded to fill the space.

Service functions absorb demand. When a new category of work arrives, they serve it, that's the operating logic they were built around. The freed hours land inside that same logic and get allocated the same way. Breaking the pattern requires something the queue never forced: a deliberate call about what the team is actually building, made before the freed hours show up and before the next wave of demand can claim them.

Service teams serve. The queue had always made that feel like enough.

The call that changes the function

This is one of the central arguments in the CIO 2026 report: freed capacity needs a designated destination before it arrives, or it flows back into demand.

Bhojwani's team worked through the identity question explicitly. The function shifted from service provider to strategic consultant, proactively identifying where IT could reduce friction across finance, HR, and operations, before anyone filed a ticket.

"Now the power is back to your individual teams," Bhojwani says. "We are acting as a consultant for them."

That shift doesn't require a reorganization. It requires the IT leader to answer a question most never face directly: what would this team be doing if no tickets came in tomorrow?

Narayan's answer at GitLab is concrete. The function moves toward building infrastructure that lets other teams operate faster, and toward overseeing the AI agents that now handle what the team used to handle manually. Same domain knowledge. A fundamentally different use of the hours. The junior hire who would have spent day one on help desk is now overseeing a fleet of agents, reviewing escalations, tuning configurations, catching the edge cases the agent misclassifies.

The job description changed. The domain knowledge remained.

One project, before the capacity arrives

Only 20 percent of organizations have AI fully embedded across service management teams, according to the State of AI in IT 2026. The gap is partly deployment pace and partly something harder to measure: how many IT leaders are still waiting for a structural mandate that will not arrive on its own.

Piru Chheang, Head of IT at Zip, ran a team of one and a half serving over 800 people. When AI deflected upwards of 50 percent of his ticket volume, he had already decided where the hours were going. A two-year-deferred offboarding automation project, work his team had started and restarted without ever finishing, because incoming tickets kept pulling them away. The freed capacity went directly to it. It shipped in a single quarter.

The function stayed in IT. What the function built changed entirely.

The org chart tends to follow the decision. It rarely leads it.

The practical starting point is narrow. One role. One project. One explicit reallocation of hours toward building something rather than resolving something. Bhojwani's framing is useful here: the shift from service provider to consultant doesn't happen via title change or a reorg. It happens when IT leadership makes a call about what the function is for, protects the hours to act on it, and lets the work change how the team sees itself.

The queue will never make that call. It will only keep answering it.

The IT's Time to Build report goes deeper on every layer of this framework, with insights from IT leaders at Lyft, Cloudflare, GitLab, Zip, and more.

Download the report.

Frequently asked questions

What does it mean for IT to shift from admin to engineer?

The orientation changes from resolving incoming demand to building systems that reduce demand and return time to the business. The domain knowledge stays the same; how the hours are spent changes.

How do IT leaders prepare their teams for this transition?

Make the decision before freed capacity arrives. Protect one project that builds leverage rather than closes tickets. The shift accumulates through completed work, not reorganization.

What happens to IT teams that don't make this shift?

Freed hours get absorbed by new demand: AI enablement requests, tool evaluations, compliance work. Without a deliberate reallocation, the operating model expands rather than evolves.

How does AI change IT hiring and role definition?

The roles shifting are toward systems oversight, agent management, and platform engineering. IT leaders who recognize this are rewriting job expectations before the next hiring cycle, not after it.

Is this realistic for small IT teams?

Yes. Piru Chheang at Zip runs IT for over 800 people with a team of one and a half. When AI deflected 50 percent of ticket volume, he directed those hours toward a deferred automation project and shipped it in one quarter.