Hello again, colleagues. Your old friend, the IT editor, is back. Today, a curious vacancy from Luxembourg has landed on my virtual desk. Victor Buck Services is looking for an experienced system administrator for Microsoft products. Well, let’s pour ourselves some coffee and see if they truly need another person on staff, or if their problems can be solved more elegantly.
Victor Buck Services is not a garage startup. It’s a serious company, a subsidiary of the POST group, involved in the creation, production, and distribution of physical and digital documents. Sounds solid and, I dare say, quite conservative. Working with documents, especially in European jurisdiction, is always about security, compliance, and reliability. Mistakes here are costly.
Judging by the job description, the company finds itself in a situation very familiar to many of us. They call it “a new adventure in M365.” I would call it “a massive and painful cloud migration with a bunch of legacy systems.” They need someone to simultaneously support the old world (Active Directory, Exchange, SCCM) and help build the new (Entra ID, M365, Intune, Purview). The pain is classic: the team is bogged down in operations. Daily checks, incident handling, patch installation, monitoring – all this routine eats up time that is so desperately needed for strategic migration tasks. They are looking for another pair of hands to put out fires and ensure nothing falls apart.
Now, let’s imagine a dialogue in their meeting room. On one side, a hypothetical IT director, let’s call him Mark. On the other, me, in the role of an invited consultant.
Mark: We need a person. Experienced, responsible. To take over the current Microsoft infrastructure routine. Our M365 project is burning, and the team is either resetting passwords or dealing with patches.
Me: Mark, are you sure you need a person? You’re looking for an employee who will execute a set of algorithms: check backups, install updates, react to alerts, close tickets. A person will do this 8 hours a day, then go home. They can get sick, go on vacation, or simply get tired and make a mistake. What if you hire a digital assistant to do the same thing, 24/7, without errors or fatigue?
Mark: Sounds like fantasy. Who will make decisions? Who will deal with unusual situations?
Me: And that’s where the whole point lies. You’re not hiring a replacement for the entire team, but a tireless junior specialist. Let’s break it down point by point from your own vacancy.
“Daily administration, handling incoming requests and incidents.” Instead of a person, we implement an AIOps (AI for IT Operations) platform, integrated with your ticketing system. A simple example: a user complains about slow email. The AI assistant analyzes Exchange, network, and endpoint device metrics in real-time. It sees an anomaly, correlates it with recent changes and the knowledge base. In 80% of cases, it either applies the fix itself (e.g., restarts a service) or provides your engineer with a ready report: “The problem is here, the probable cause is this, the recommended action is this.” Reaction time is reduced from hours to minutes.
“Performance tuning, software updates, patching.” You already use Ansible. Excellent! Let’s give it “brains.” An AI monitoring system (e.g., Datadog or Dynatrace, and Microsoft has its own solutions in Azure Monitor) doesn’t just show graphs. It predicts problems. “Disk on server X will be full in 48 hours.” “Load on domain controller Y abnormally increased after patch Z.” This same system, seeing a new security bulletin, can itself assess the criticality of the vulnerability for your infrastructure, check systems for compatibility, and schedule patch installation via Ansible during the lowest load window. A person only confirms the operation.
“Regular security monitoring.” Colleagues, this is the pure domain of AI. Modern EDR and SIEM systems (like Microsoft Sentinel, which would fit perfectly into your M365 ecosystem) are no longer about signature-based searches. They’re about behavioral analysis. AI learns how your systems and users “normally” behave and raises an alarm at the slightest deviation. A human administrator is physically incapable of tracking thousands of events per second. AI can.
How to overcome distrust? Start small. You don’t need to immediately give AI the keys to everything.
Step 1: “Advisor” mode. Launch the system in monitoring mode. Let it simply analyze and provide recommendations. Your team will see its conclusions, compare them with their experience, and gradually become convinced of its competence.
Step 2: Risk-free automation. Entrust AI with the most routine and safe tasks. Checking backup status, cleaning temporary files, monitoring certificate expiration. The team will see how much time this frees up.
Step 3: Controlled execution. Allow AI to perform more complex tasks, but with mandatory human confirmation. For example, “I found a critical patch for 50 servers, the playbook is ready. Do you authorize installation?”
And how to verify that all this works? Very simply. The result is validated not by emotions, but by metrics.
Firstly, classic IT indicators: Mean Time To Resolution (MTTR), percentage of automatically resolved tickets, system downtime. These should radically improve.
Secondly, team workload. Conduct a survey before and after implementation. How much time do engineers spend on routine, and how much on that strategic M365 project? The result will surprise you.
Thirdly, business indicators. Reduced security risks, increased user satisfaction (because their problems are solved faster). This is something even a financial director will understand.
Instead of looking for another “one-man band” who will be torn between the old and the new, Victor Buck Services could invest in an intelligent assistant. This would allow their current, already proven team, to lift their heads from operations and focus on the architecture of the future. And that, as we know, is the most interesting and valuable work in IT.
Источник: https://www.linkedin.com/jobs/view/4414007538/
