How artificial intelligence is changing the daily work of a senior IT consultant — and what it does not change.
A few years ago, AI in IT mostly meant marketing slides and chatbots that could not answer a single real question. Today it is part of my working day. Not as a revolution that replaces consultants, but as a tool that quietly changes how I research, write, script, troubleshoot and communicate. In this post I want to share how AI shows up in my daily work around Microsoft 365, Azure Virtual Desktop, Intune and automation — and where I draw the line.
From searching to asking
Much of an IT consultant’s day used to be spent searching: documentation, forum threads, release notes, GitHub issues. The answer was usually out there, but finding it took time. With AI assistants, the first step has shifted from searching to asking. Instead of opening ten browser tabs to understand why an FSLogix profile container fails to attach, I describe the symptoms, the event log entries and the environment — and get a structured list of likely causes to check.
That does not make the documentation obsolete. It changes the order: AI gives me a hypothesis and a direction, and the official sources confirm or correct it. The time from “something is broken” to “I know where to look” has become noticeably shorter.
Scripting and automation: faster drafts, same responsibility
Automation has always been at the core of my work — PowerShell, Microsoft Graph, Infrastructure as Code and deployment pipelines. This is where AI has had the most visible impact. Boilerplate code, parameter validation, error handling, Graph API calls with paging, or a first draft of a remediation script for Intune: an assistant produces a solid starting point in seconds.
But a draft is not a deliverable. Generated code regularly uses outdated cmdlets, invents parameters or ignores edge cases that matter in production — throttling, permissions, multi-tenant setups or a device that has been offline for weeks. My role has shifted from writing every line myself to reviewing, testing and hardening what is generated. The engineering skill is still required; it is just applied at a different point in the process.
Troubleshooting and incident response
In incident situations, time and clarity matter. AI helps me structure the chaos: summarizing long log excerpts, correlating timestamps across sources, suggesting KQL queries for Log Analytics or Defender, and turning scattered notes into a clean timeline. When a user session on Azure Virtual Desktop hangs at logon, I can move quickly from raw data to a prioritized list of things to check.
What AI cannot do is understand the specific environment the way someone who designed or operates it does: the historically grown group policies, the exception that was made for one department, the change that went live last night. Context remains the consultant’s job — and often it is the decisive factor.
Documentation and communication
A large part of consulting is communication: concepts, architecture decisions, runbooks, status updates, workshop summaries, release notes. This is an area where AI saves a lot of time. I can turn bullet points into a well-structured concept, adapt a technical explanation for management, or produce a first version of a runbook from a working session.
The result is not only faster documentation but often better documentation — because the hurdle to write it down at all becomes lower. Still, I review every sentence that goes to a customer. Tone, accuracy and accountability remain mine.
AI in the Microsoft ecosystem itself
AI is not only a tool for consultants; it is becoming part of the platforms we design. Microsoft 365 Copilot, Security Copilot and AI features in Intune and Defender change what customers expect from their workplace and their operations teams. This brings new consulting questions:
- Is the data governance ready? Copilot surfaces what users can access — including what they should not have access to.
- How are permissions, sensitivity labels and SharePoint oversharing handled before rollout?
- Which use cases create real value, and which are just licenses nobody uses?
- How do operations teams integrate AI-supported analysis into their existing processes?
In many projects, the most important AI work is not the AI itself, but the groundwork: identity, permissions, data classification and clean structures. Zero Trust and good governance are prerequisites, not optional extras.
Where I draw the line
Using AI responsibly in a customer context requires clear rules. For me, these are the most important ones:
- No sensitive customer data in public tools. Tenant IDs, user data, configurations and logs are anonymized or only processed in approved environments.
- Every result is verified. AI output is a draft and a hypothesis, never a final answer.
- Accountability stays with me. A customer does not care whether a script was written by hand or generated — they care that it works and is safe.
- Understanding beats copying. If I cannot explain why a solution works, it does not go into production.
What changes — and what stays
AI makes me faster at the parts of my work that are repetitive, text-heavy or search-intensive. That frees up time for what actually creates value for customers: understanding their situation, designing sound architectures, making trade-offs, and taking responsibility for the result.
The role of the senior consultant is shifting from being the person who knows every answer to being the person who asks the right questions, evaluates answers critically and turns them into solutions that work in the real world. Experience does not lose value through AI — it becomes the filter that separates a plausible answer from a correct one.
How is AI changing your daily IT work? I would be happy to exchange experiences.