“There are no fast followers for AI.” That was the message from Eduardo Kassner, Microsoft’s Chief Data and AI Officer, at Extreme Networks’ Connect 2025 event in Paris. It stuck with me because it is true. You do not get to wait this one out. By the time you have decided to jump in, those who moved early will already be far ahead.
I saw that reality play out live on the main stage. Nabil Bukhari, Extreme’s Chief Technology and Product Officer, and Hardik Ajmera, Vice President of Product Management, ran a troubleshooting demo using AI. It took a complex performance issue, something that could have taken a skilled engineer hours to diagnose, and resolved it in minutes. The AI handled most of the steps automatically, surfacing clear actions without drowning the operator in guesswork or logs.
As a CTO, I found the capability impressive. But I also noticed something else. While some attendees were visibly excited, others were cautious. More than a few network professionals told me they were uneasy about relying on AI for something as critical as network operations.
I understand that hesitation. Accuracy and trust matter. The idea of letting an AI system make calls about your network feels like a leap. But AI already makes fewer mistakes than people, and it is improving every single day.
Complexity is our normal
In South Africa, we live in a world of “complicated Wi-Fi,” whether we call it that or not. Our clients run networks in places that were never designed for high-density wireless. Thick brick walls, steel roofs, and unpredictable plans. We contend with ongoing electricity interruptions, limited budgets, and high RF interference.
In those conditions, troubleshooting is never straightforward. You are often piecing together clues from different systems, walking between buildings, waiting for the right people to be available, or trying to replicate a problem that happened hours ago. It is slow, human-intensive work. And while it is happening, your users are still experiencing poor connectivity.
AI changes that equation.
Instead of waiting for the helpdesk ticket or walking into a room full of frustrated users, AI can spot the problem before they even notice it. It can tell you which access point is behaving oddly, identify the affected devices, and even suggest the most likely cause. And because it learns from every incident across the network, it becomes better at recognizing those patterns over time.
From firefighting to foresight
The real breakthrough is not just speed, but foresight. AI-enabled platforms like Extreme’s can predict issues before they impact the business. Think about what that means for a hospital relying on wireless patient monitoring or a logistics hub tracking inventory in real time. It means downtime is avoided, reputations are protected, and IT teams are freed from constant firefighting.
It also means you can spend more time on meaningful projects. AI is not here to replace engineers. Rather, it is here to give them back the hours they lose to repetitive, reactive work. And in an industry where skilled networking staff are in short supply, that time is invaluable.
Lessons from Paris and from home
One of my favorite case studies at the event was Anfield Stadium in Liverpool. It is a 139-year-old structure, retrofitted with high-performance Wi-Fi without compromising the building’s heritage. That kind of challenge felt familiar.
Here in South Africa, we have our own “Anfields.” For example, historic schools, government buildings, and commercial parks that need modern connectivity without major structural changes.
AI fits perfectly in those contexts. It makes the most of the network you have, squeezing out performance gains, spotting hidden issues, and keeping things stable in environments that are anything but simple.
We have started applying these principles locally. On large urban campuses and in remote warehouses alike, we are using AI to design and manage networks that deliver consistent performance. We are layering in LTE fallback for resilience, enabling mesh-aware routing for coverage, and letting cloud-managed AI do the heavy lifting on monitoring and optimisation.
The call to action
If you are a network professional in South Africa, I believe now is the time to lean into AI. The complexity is already here. The budgets and headcounts are not getting bigger. What can change is how you use your time and how quickly you can act when the network is at risk.
Waiting until AI in networking feels “safe” will not just mean you are behind on the technology. It just means you are behind on skills, efficiency, and on delivering the experience your users expect.