Key Points
- AI in IT support improves efficiency by automating repetitive Tier 1 tasks, reducing manual workloads, and helping technicians resolve issues faster.
- Tier 1 support automation can struggle when users provide incomplete, inaccurate, or emotionally driven descriptions of technical problems.
- Human support staff add context, empathy, and problem-solving skills that AI systems cannot consistently replicate during stressful support interactions.
- Human-in-the-loop IT support combines AI-driven pattern recognition and knowledge retrieval with human judgment and communication.
- High-performing support organizations use AI to enhance service delivery while maintaining human relationships that improve customer satisfaction and trust.
The IT industry is racing to automate Tier 1 support. Gartner found that 91% of customer service leaders feel pressure to use AI this year to improve operational efficiency. Another study by McKinsey showed that 64% of enterprise leaders say that AI is enabling their innovation, and 80% list efficiency as the primary goal for all AI projects.
However, AI in IT support was never really about replacing humans. It’s always been about helping them do their jobs better. Companies that succeed with AI aren’t asking “How do we automate our support desk?” Instead, they’re asking “Is our Tier 1 support actually doing what it should?”
IT leaders see these numbers and immediately view Tier 1 support as the most logical first step for automation. After all, this is only the first point of contact, right?
So, what’s the issue?
Tier 1 support automation isn’t science fiction
The person submitting a ticket is often stressed and may not know how to explain what’s wrong. This is where automation can fall short.
Even the best AI models have trouble reading context. Their job is to solve the problem, but this gets harder if the user can’t explain what’s happening.
For example, if you visit a doctor for unknown pain in your chest, you wouldn’t say, “I am experiencing sharp left-sided posterior chest pain that worsens with deep breaths.” Most people would just say, “It hurts when I take a deep breath.”
The same applies to AI in IT support. Frustrated users may not describe the problem accurately, and AI can only work with the information it gets, often depending on how the prompt is written. Humans, though, can read between the lines and ask for more details.
Human-in-the-loop IT support
Research in adjacent fields has found that AI-administered assessments are poor predictors of real-world outcomes. This is because AI can only operate with what it is given.
Ask AI how to solve a problem, and it will come up with an answer. This is great – until you need to add nuance and emotion.
Think of the last time you called IT support. More than likely, you were stressed, and that affects your communication skills. In times of stress, people tend to simplify language, and it may not always be accurate. Here’s an example: Your laptop died five minutes before a big meeting. You’re definitely not in the proper headspace to file a precise, well-reasoned ticket. You drop context, you misdescribe symptoms, you say “It’s broken” when you really mean “I don’t know what just happened. Help me. My laptop went into cardiac arrest.” Either way, you can’t fully communicate exactly what went wrong (and that’s perfectly normal).
In Tier 1 support automation, this can lead to imprecise outcomes. If a user, as their first point of contact, is given a list of topics to scroll through or describe their issue to an AI bot, the list may not contain the questions needed to describe what the user needs.
Your IT automation, though, learns through consistent interactions. AI models are excellent at pattern recognition and can be trained to offer “better” topics in the future. In fact, it can even be argued that we are training ourselves to communicate better with AI, rather than the other way around.
Why AI shouldn’t replace IT support staff
Tier 1 support staff are called the “frontliners” because they are the first people users talk to when something goes wrong with their devices or software. When they’re experiencing technical difficulties, can we really expect users to have the patience or understanding to deal with an AI-powered IT help desk?
Research shows that most people still prefer talking to a human, even when they could get answers through automated systems. McKinsey’s own data reframes the conversation, stating that the question was never “AI vs. human support,” but how AI can support humans.
The better approach is to support your IT help desk technicians with AI that handles knowledge bases and spot patterns, so support staff can focus on helping the person on the other side of the ticket.
As Michael Shelton, VP of Global Product Support, puts it: “Our relationship begins with our first contact. We strive to ensure that every interaction is easy, refreshing, and successful.” The result? A 98.4 CSAT (customer satisfaction) score and a 31-minute average response time, not because we automated humans out of the equation, but because we gave them better tools and let them do what they do best.
That’s what AI in IT support should look like.
