After my last BNI meeting, I stayed behind talking with a fellow member, and he told me how much AI had helped him with his IT questions. He had bought a business-grade firewall, and rather than calling anyone, he asked an AI tool to walk him through the setup. It did. He still had a couple of questions, one of them about getting two pieces of networked equipment to work together, but for the most part the AI got him up and running. Longtime readers know I consider that the right purchase, since the box your internet provider hands you is not a firewall.
Then he turned the conversation around. He wanted to know how I use AI in my own business, and, more pointedly, how AI has changed the services we provide to clients. It was a fair question, coming from someone who had just finished an IT project without calling anyone. It is also the question I hear most often about AI and managed IT.
I will start with the help desk, because that is where people assume AI has done the most damage to companies like mine. For years, much of what people valued in managed IT was someone to call when Outlook misbehaved or a printer vanished from the network. AI now answers a good share of those how-to questions, and for the more technically inclined, some complicated ones as well. We are happy when clients use it to learn how to do a task. But the phones have not gone quiet. The calls have changed. More and more, the question is not “how do I do this?” but “I read this, or AI told me this. Is it accurate, and is it safe?” That reassurance is something a chatbot cannot credibly offer about its own answer. And every hour an employee spends troubleshooting their own tools is an hour not spent on the work you hired them to do.
Where AI and Managed IT Work Together
Here is how we use AI ourselves. It does something for us that we could never do well before: it watches. We run AI-driven monitoring on computers, on networks, and on email accounts, around the clock. It is a guard dog that never needs to eat, sleep, or step outside. When it barks, every alert goes to a person on our team to investigate and, when necessary, to mitigate. The AI notices, and a human decides. According to IBM’s 2026 Cost of a Data Breach Report, companies using AI and automation in security operations cut breach costs by an average of almost two million dollars, and phishing led all breach entry points for a fourth consecutive year.
Where AI Falls Short on Security
We have also learned where it falls short. AI still makes remarkably simple mistakes, delivered with complete confidence. A few months ago I wrote about asking an AI tool to count the letter e in the word seventeen, which took four attempts. More subtly, AI will answer the question you asked without mentioning the ramifications of the answer. Ask how to get two devices talking to each other, and the fastest route to a working answer may be loosening a rule that existed for a reason. The devices connect, the problem appears solved, and nothing in the answer tells you what you just opened up. I have no reason to think that happened to my BNI friend. My point is that the tool aims for “it works,” and “it works” and “it is secure” are different questions.
That distinction matters because security is almost always a negotiation between what is safest and what people will tolerate. Every extra login prompt, blocked download, or locked-down setting buys protection at the price of convenience. Deciding where to draw that line for a particular business, with particular people and particular data, is a judgment call, and a tool answering one question at a time is not positioned to make it.
What Managed IT Services Still Do
So here is what has changed. The help desk is still busy; the questions are simply different, and the work behind it matters more than ever. Someone still has to manage the security dashboards, configure the tools correctly in the first place, keep firmware and software current, and, most importantly, know which alerts demand immediate action and which ones are noise. Email is the current battleground, and it is where that judgment gets tested every day.
Managed IT cannot prevent every problem, and anyone who promises otherwise is selling something. What it can do is reduce your exposure and, when something does go wrong, make sure you can recover. In regulated industries, meeting the security requirements is not optional, although who does the work is. For everyone else, it is a choice, and a reasonable one to weigh on its merits.
If I had to boil my answer down, it would be this: AI changed what our clients need us for. It did not change whether they need someone watching.
Until next time, keep IT simple.
Dave
CMIT Solutions of Wilmington
