By Andrew Beynon, Chief Enterprise Architect.
For those of you who missed Ed Meyercord, President and CEO of Extreme Networks, CRN interview on Platform ONE and its potential, and my follow-up with Gina Narcisi in CRN’s 29/9/26 article, Read Here: for many years, network management has largely been about understanding infrastructure.
Is the switch healthy? Is the access point overloaded? Is the WAN link experiencing packet loss? Is an application slow? Is a client having connectivity problems? So on and so on.
However, with the explosive adoption of Artificial intelligence, vendors are fast beginning to change that model. Extreme Networks’ aggressive AI strategy and arguably the pinnacle of engineering product Platform ONE, is challenging the competition and transforming how we view, utilise, consume and manage both users and network services.
The more interesting opportunity is not simply to use AI to find problems within the network. As I said in the CRN segment, there is no shortage of AI and no shortage of network tools; what makes Platform ONE different for me is that it is built on AI, not having AI bolted on as an afterthought.
I will soon travel to Amsterdam to represent ARO at the Extreme Networks AI Summit 2026, where I’ve been invited to join a panel of experts, along with others from select organisations, to look beyond the hype and share insights on transforming AI from a vision into a genuine network advantage. Now it’s about imagining the possible and leveraging the toolsets at our disposal.
I have long said the network is a source of behavioural intelligence, and if we can understand how users, devices, applications, locations and security controls normally interact, and then identify when those relationships change, it opens a huge potential not just for better security, but ultimately a better client experience.
Extreme already brings together much of the required telemetry. Its analytics capabilities provide visibility of applications, users, devices and locations, while application telemetry can provide information about application performance and network flows, and Platform ONE brings together wired, wireless and SD-WAN management, client health and application visibility, while machine learning and anomaly detection demonstrate network behaviour.
Today’s conversation is no longer focused on “Is the network working?” It’s changed to “Is this behaviour normal?”
If you consider a typical employee, whatever typical looks like, and by leveraging the information over several weeks, we could establish a behavioural baseline for that user’s normal working pattern. I don’t mean simply how much data they consume, but which devices they normally use. Which access points and locations do they normally connect from? Which applications do they normally access? At what times do they normally work? Or even which SaaS services do they normally use, what is the normal application latency, and so on and so on.
Couple this insight with which network segments they normally access? Which authentication and access policies normally apply? How frequently they move between locations. We could then quickly establish a vast degree of information about the client and device, and whilst individually, none of these signals necessarily means anything.
Collectively, however, they create a behavioural fingerprint and digital footprint. For me, the power of Platform ONE, Extreme’s Universal Zero Trust Network portfolio, and wired and wireless products offer huge potential because of the significance of AI being the foundation and guiding light of the product set.
Aside from the fact that we would no longer need to manually establish a persona, as this fingerprint does not need to be manually defined by an administrator, I believe this demonstrates the potential of the platform for long into the future.
Being able to craft a solution which collects, collates, and classifies not just a user but every user’s foot and fingerprints will allow us to establish a dynamic baseline and continually compare new activity against it, I think this will transform how we manage networks and user experience moving forward.
Imagine an employee who normally works from Cardiff, who normally connects using a corporate laptop, accesses Microsoft 365, Teams, a small number of internal applications and several known SaaS platforms, normally connecting to a Wi-Fi 7 app.
Then one morning, the system observes something different. The user authenticates from a new location, the device is unfamiliar, and the user connects to a different wireless infrastructure, and their application profile changes significantly, which now shows there is an unusual increase in outbound traffic!
They begin accessing applications they have never previously used; their traffic pattern changes from predominantly business applications to unfamiliar external services.
Whilst none of those events necessarily constitutes a security incident, because the user may actually be at a client site, that may not be the case, and this could be a looming issue or soon become one.
I believe we are now at a reflection point: do we now trust our gut instincts over the technology? But AI does not have to treat them independently, nor does it have to treat them differently. This will be AI adoption at the speed of trust.
There is a much more powerful question: “Is this individual event anomalous?” or is the correlation where the real value emerges?
I foresee a point where the power of the AI engine could continuously correlate these dimensions, process both current and historical information and quickly establish a pattern of not just security concerns but support challenges; for example, if the remote site is blocking traffic or applying restrictions to the user, having context awareness simultaneously changes the probability that the behaviour is meaningful.
So in this case the user’s current behaviour deviates significantly from their established behavioural baseline; the important point is that this does not necessarily mean the user is compromised. It means the behaviour warrants investigation.
That distinction matters. I think Extreme’s coworker and future announcements will soon shift the focus, moving these aspirations from whether a person should identify the anomaly and explain the evidence behind it to what AI thinks the organisation should determine and do as the appropriate response; possibly even doing it for you.
As I said at the start of this post, Extreme Platform ONE, I believe, allows aspirations from partners like ARO and my vision of the future to become a reality, moving beyond the hype to create an AI-driven contextual intelligence layer that understands the relationship between people, devices, applications, infrastructure, security and experience. Again, the question is: is it AI for networking or networking for AI?
The differentiator I feel is therefore less about having another AI assistant and more about having an AI engine with access to an unusually rich network context that works with you, as well as for you, and that is an important distinction.
The most interesting evolution of AI networking may not actually be autonomous configuration and self-driving networks. It may be autonomous understanding, which could develop into a continuously evolving understanding of what normal looks like for every user, device, application and location.
Watch this space. I am sure the AI Summit will answer as many questions for us as it poses.
By attending and participating in events like this, it enables ARO to see networking differently. The network is not simply the infrastructure that moves data between endpoints; it is the connective fabric that brings together people, applications, data and technology to enable organisations to achieve their ambitions.
