The future of IT operations is humans + AI (not either/or)
.png)
Table of contents
The seismic shift in how IT work gets done
In 1698, when Thomas Savery arrived at an English mine with the world’s first operational steam engine, most of the miners probably appreciated having a handy new tool to move water out of their way. What they probably didn’t think was, “We’re witnessing the beginning of a new era in human history.” But that's exactly what was happening.
Fast forward to 2025. The rise of generative AI is one of the most significant technological developments in human history. Bigger than the steam engine. Bigger even than Thomas Savery’s hair.
We’re now watching AI remake how we work — in real time. And nowhere is that transformation more urgent (or more overlooked) than in the world of IT operations.
For decades, IT teams have operated in a tug-of-war between quality and scale. Deliver personalized, white-glove support? Great, but expensive. Scale to serve a growing business? Sure, but users might feel like they’re just another ticket in the queue.
This trade-off has only gotten more painful. With companies now averaging over 150 SaaS apps, hybrid teams spread across time zones, and digital-native employees demanding lightning-fast, consumer-grade experiences.
The traditional IT help desk model is hitting a wall.
Designing systems where AI agents, automation, and humans work together to deliver support that is empathetic, efficient and endlessly scalable will change the way IT works.
At Fixify, we believe the natural outcome of an overly-relentless focus on automation and efficiency is … alienation.
We think the admixture of human craft and automation (including AI) creates a richer, more effective, experience.
In short, the answer isn’t to choose between humans or automation — it’s to design systems where AI agents, automation, and humans work together to deliver support that is empathetic, efficient, and endlessly scalable.
And this shift will fundamentally change the nature of work. Full stop.
What’s breaking the old model
Let’s go back to that tale-as-old-as-time tug of war between quality and scale.
Several undeniable trends and macro forces are etching bigger and bigger cracks in the way IT works — and particularly the way IT help desk work gets done.
Here are the big ones:
- The rise of remote work: More and more employees are remote. That means there’s increasingly no actual relationship with IT in the office. And it’s not just IT where relationships have worn thin. By some reports 18% of the workforce is actively disengaged.
- SaaS app sprawl: The number of SaaS tools has exploded to 150+ per company. The train left the "here's the office suite you're using" station long ago. Increasingly apps are owned and administered by each department (not IT). IT is increasingly a facilitator — not necessarily the do-er when users file tickets asking for access to this, that and another app.
- New generational expectations: Younger generations of employees have new and different expectations for what good service looks like and how they want to interact with IT. 60% of Gen Z prefers self-service support.
- BYOD → BYOT: It’s not just “bring your own device” anymore. Increasingly people are bringing their own tools. Can’t get access to the company’s preferred AI note-taker? No problem, they’ll grab their own. Same with ChatGPT or Claude.
This all lands hardest on IT teams at growing companies. A 300-person company might have just three IT staff, responsible for 150+ tools across every function. And many of those tools don’t integrate cleanly — or at all.
As one IT exec told us, “I click as fast as I can all day, and I still can’t make progress.” That’s what burnout sounds like.
All of these trends are keeping too many IT teams stuck in reactive mode. Drowning in tickets. Swamped by provisioning. Blocking on strategic projects they never have time to start.
And here’s the kicker: when end users are unhappy, no dashboard can save you. That’s the new bar for IT — keep users happy or lose trust. And to hit that bar, something fundamental has to change.
A bold future for IT
When I talk with CIOs and VPs of IT about what's actually working at companies where people love their IT teams, I hear things like: "we put in tech vending machines" or "we put in a walk-up bar." These are in-person experiences with a particular texture designed to evoke specific reactions. They’re not really focused on efficiency — they’re focused on choice, empowerment, and maintaining a personal touch. These are great qualities. But they’re hard to scale.
We need to change the way we work.
While the nuts and bolts of how IT work gets done will indeed evolve and change (a lot) over the next ten years. The mindset will also change.
Here are the four big shifts I see coming to IT in the years ahead.
| FROM | TO | |
|---|---|---|
| Automation |
→ |
Personalization
|
| Closing tickets |
→ |
Care and craft
|
| IT as a last resort |
→ |
IT as the first stop |
| Reactive |
→ |
Predictive |
| Hurdle | Why it's a hurdle | How it gets in the way |
|---|---|---|
| Variation | Since IT exists to support the org (not vice versa) every IT environment is unique. There’s a different mix of tools and the processes are a little different. |
Variability is the enemy of automation. “Out-of-the-box” tools become hard to find. Variability leads to either A LOT of customization or else you’re left automating the subset of the problem where there’s less variability.
|
|
Edge cases
|
Even in orgs with consistent, well-documented processes the common case inevitably falls victim to uncommon circumstances. The VIPs don’t file tickets. An application upgrade fails because of a one-off driver issue. Pick your poison. |
AI does reasonably well if it’s been trained on the right data. But edge cases by definition, aren’t in the data. And that’s where AI taps out and you need old fashioned humans to tag in.
|
| Human/ tool interface | Two words: “CSV export”. The state of integration between IT tools is poor. It’s hard to visualize processes and harder to manipulate them. | Without an easy way to visualize a process it’s hard to automate it. When you add in AI agents, if you can’t understand what they’re doing behind the scenes (or why) it’s hard to trust them. That slows adoption. |
| Technology sprawl |
With SaaS apps multiplying like rabbits, every person in IT is stretched a mile wide and an inch deep. Plus, increasingly those SaaS apps are owned by each function and IT is a facilitator (not always an owner).
|
Increasingly IT teams don’t know enough about each app to automate a process or point AI at the problem. And they don’t have time to climb the learning curve. |
| Bad user experience | There’s that old adage: “technology would be great, if it weren’t for the people using it.” There’s truth in the fact that adoption is only as good as the quality of the experience. | Not everyone wants to talk to a bot. And when people can’t self-service an answer to their problem out of the AI ticket-deflection tools they're likely to skip it the next time they have a problem — dragging down adoption rates. |
The new model: humans + AI agents working together
So if the way work gets done is changing, what does that actually look like in the real world?
Right now, most IT teams are doing one of two things:
- Building manually coded rules-based automations: This includes if-then-style approval flows that are manually coded or created with tools like Okta Workflows.
- Using AI to support humans (or defect tickets): This includes tools that act as copilots for IT analysts or attempt to help users help themselves. Think: ChatGPT-style assistants that draft replies, virtual agents like FreshWorks’ “Freddy” that handle basic triage, or platforms like Moveworks that try to infer intent from a few keywords and point users to documentation. These tools can lighten the load — but they often struggle when context is missing or knowledge is siloed.
These are natural starting points. But the future isn’t “AI as assistant” or “automation as savior.” It’s a new model where AI agents, automation, and human expertise operate in sync.
There are three ways AI agents will help perform IT work. Here’s what each of them looks like.
AI agents working alone
In this model, AI agents independently handle well-defined, repetitive tasks that follow a consistent pattern and don’t require human judgment. These tasks are the bread-and-butter of many help desks — simple, high-volume requests that clog up queues but don’t need escalation or nuanced thinking.
Think of these agents as highly efficient junior analysts who never sleep, don’t get distracted, and never forget the steps in a runbook. They can follow instructions, make basic decisions based on structured inputs, and take actions across systems using APIs. Importantly, they don’t need a human in the loop to finish the job.
Examples:
- Password reset: When a user locks themselves out of their account, an AI agent automatically resets credentials after verifying identity using contextual data from the user’s SSO and device history. No human needs to touch the ticket.
- Onboarding: A new hire is onboarded, and the AI agent provisions accounts across 12 core SaaS tools — grabbing the right license tier from Salesforce, assigning the right Google Workspace group, and enrolling them in the company’s MFA policy in Okta.
- Printer down: A “printer not working” ticket comes in. The agent identifies the printer model and user location from the ticket metadata, pushes a driver reinstall command to the device via an RMM tool, and verifies resolution by checking for a successful test print.
These tasks aren’t flashy, but they represent a meaningful share of help desk volume — and a massive opportunity to reclaim time and improve consistency.
AI agents collaborating
This model is like the brooms in the Sorcerer's Apprentice, if Mickey Mouse had read to the end of the instructions and it hadn’t all gone to hell. A team of complex, specific agents work together to solve a multi-part problem that spans several systems, stakeholders, or approval paths. These aren’t tasks that can be knocked out with a single command — they require coordination, sequencing, and adaptive logic depending on domain-specific context.
The key here is orchestration – one agent fetches the right data, another applies policy rules, another triggers the necessary actions, and yet another confirms everything was completed successfully. This unlocks the ability to scale workflows that used to require multiple humans and handoffs.
Examples:
- Onboarding: A new sales hire in London triggers a workflow where one agent verifies role and region, another provisions Salesforce access (this part might take a whole team, tbh), a third adjusts calendar and communication settings, and a fourth sends a Slack welcome with key resources. Each step is personalized based on team, title, and location.
- Offboarding: When an employee leaves, agents deactivate accounts, archive and transfer Google Workspace data, revoke group access in Okta, and alert HR and Legal — all in sequence, with built-in checks for security risks or leftover sessions.
- Contractor access: A contractor’s access request kicks off a workflow where agents verify contract dates, grant time-limited access to Netsuite and Xero, log activity, and schedule a review — automating a process that used to involve multiple emails and approvals.
In all of these workflows, we might add a quality assurance agent that checks the results against known good for that type of user and workflow.
These collaborative agent workflows are especially powerful in environments with lots of SaaS sprawl and policy complexity. They’re still only handling the common-case paths, but they can handle slightly more inter-domain complexity than the single-agent scenario.
AI agents collaborating with humans
As we see it, this is not an evolutionary step, it’s the destination. Multiple AI agents work in tandem with humans to achieve complex results that challenge the limits of both. This is both Jarvis and Tony Stark, working together to solve the problem of time travel.
The task here isn’t mechanistically dispatched to a team of agents – rather work flows organically through a human / AI hybrid team, where we get the best of both – the fantastical breadth of AI combined with the ineffable human quality of judgement.
Examples:
- Contextual prep: An agent sees a ticket come in from a frustrated exec and instantly pulls a timeline of recent issues, device diagnostics, and prior interactions — giving the analyst a clear view of what’s happening before they even respond.
- Sentiment-aware escalation: A user replies with “this is ridiculous, I’ve already done that,” and the agent flags the emotional tone, routes the ticket to a senior analyst, and recommends a more personalized, empathetic response — avoiding a potential trust breakdown.
- Smart handoff: During a software troubleshooting request, the agent handles 80% of the diagnostic steps — collecting logs, validating environment details, running standard scripts — then hands off to a human when deeper investigation is needed. The analyst picks up with full context and zero duplication of effort.
These AI-human tag teams are especially effective in messy or emotionally charged situations, where accuracy and empathy matter. The agent lifts the load, the human brings the nuance — and together, they deliver an experience that’s fast, thoughtful, and high trust.
Where we go from here
Let’s be real: this transformation won’t happen overnight.
The IT help desk isn’t going to magically turn into a revenue engine next Tuesday. But the shift is already in motion — and it’s happening faster than most people realize.
And if we zoom out, the ripple effects could be nothing short of revolutionary.
Think back to that steam engine Thomas Savery wheeled into an English mine in 1698. At the time, it was just a clever tool to move water. That was the first-order effect. But the real transformation came later — when steam engines powered locomotives, which led to travel that opened up entire continents and reshaped economies. Those were the second-order effects. They changed everything.
We’re at a similar moment now.
Today, we’re focused on first-order wins: faster ticket resolution, fewer manual tasks, lower operational overhead. That’s good. Necessary, even.
But the real impact of AI in IT won’t just be about saving time. It’ll be about how those saved minutes compound into something bigger.
In fact, we’re just starting to imagine what those second-order effects might be. But make no mistake — they’re coming.
So I’ll leave you with this:
The companies that win over the next decade won’t be the ones with the lowest ticket counts or the flashiest bots. They’ll be the ones that master the craft of combining people and machines to deliver care at scale.
The future of IT isn’t either/or. It’s humans + AI. Let’s build it.
Related articles
What is a help desk? Definition, functions, and best practices
Is outsourcing IT support right for you?
The ultimate employee onboarding checklist: Streamlining IT setup for new hires
Stay in the loop
Sign up to get notified about our latest news and blogs