Your customers are excited about AI. But are they thinking about what it's going to do to the network underneath it?
In this episode, Viktoria Szindekovics and Dave Garwood spoke with Roy Verboeket, VP Sales Engineering at Extreme Networks, about the infrastructure challenge that tends to get overlooked in the rush to adopt AI, and what it means for how partners sell.
Key takeaways
The trend to pay attention to isn't AI: What it does to the network
Roy's opening point was that everyone's talking about AI: which model is smartest, which tools write the best emails. But at an enterprise level, the thing partners should actually be paying attention to is the network that has to support the AI transition.
A network built five years ago was designed around one assumption: humans clicking buttons. Someone opens a browser, a server responds. Predictable, manageable, human-speed. AI works completely differently. It's machine-generated, never sleeps, and runs continuously. Every time a user runs an AI query, dozens of backend transactions fire off automatically. Databases get searched, models get called, results get assembled—all in seconds. Multiply that across an entire organisation running it all day, and you can see where the pressure builds.
Roy's analogy: it's like a road perfectly designed for family cars that overnight starts carrying container trucks, 24 hours a day. The road itself isn't broken. The system just wasn't built for that.
When a technology conversation becomes a business conversation
Roy's answer was direct: "The moment you stop talking about what something does and start talking about what it costs when it doesn't."
He described a framework he uses with customers. Three questions:
· Do I make them money?
· Do I save them money?
· Do I reduce their risk?
If you can answer yes to any one of those, you have a business conversation. You just need to find that angle and go deeper on it.
He also pointed out that the person sitting across the table matters. A network admin has a different agenda from a CFO, CTO, CIO, or CISO, and the conversation needs to adjust accordingly.
The biggest AI mistake: expecting it to just work
Roy's view is that companies treat AI like something they can buy off the shelf and expect magic. The problem is that the underlying foundation of AI is data, and if the data is messy or disorganised, it doesn't matter how good the tools are.
He shared a real example. Before a recent customer meeting, he discovered the customer was a passionate home cook with what looked like five ovens in his background.
So, Roy used that.
Buying a top-of-the-line oven doesn't make you a great cook overnight. Without good ingredients, a fully functional kitchen, and someone who knows what they're doing, all you'll get is an overcooked meal. AI is the same.
He also flagged the growing importance of involving the CISO early. Governance, compliance, and guardrails need to be in place. In sectors like healthcare, there are strict rules around patient data—things like PII—and even selling a switch or AP means your infrastructure has some visibility into that. The rules need to be there.
Network visibility: the smoke detector nobody wants until they need it
Roy described visibility as consistently undervalued, right up until something goes wrong.
He and Dave agreed: they've never seen a customer cheer when visibility comes up in conversation. The value only becomes clear after something breaks and they realise they had no idea what was happening. Then, once the forensics are done, the picture emerges of what could have been prevented.
Roy's analogy: it's like a smoke detector. You can buy a cheap one or an expensive cloud-connected one that costs 100 to 200 pounds and gives you all kinds of alerts. People complain about the cost until the day it beeps, and you can put out the fire on the stove before the house burns down. That's visibility.
He connected this to what Extreme is building with Platform One and Agent One: the idea of catching issues before customers even notices them.
The shift from systems engineering to sales engineering
Roy described it as a shift that's still happening, even after all these years.
He explained the difference. A systems engineer understands the technology deeply. A sales engineer looks at it in a different light, they can move between the technical detail and the business conversation. Roy noted that someone who started out configuring networks via CLI approaches things very differently from a 25-year-old engineer today who works with APIs and can do in five minutes what used to take five hours or five days.
Three actions partners should take now
The conversation with Roy Verboeket highlights a clear opportunity for partners: help customers look beyond AI applications and focus on the foundations needed to deliver successful outcomes.
Assess customers' network readiness for AI
AI places new demands on networks, security, and operations. Yet many organisations are focused on the application layer without fully understanding whether their infrastructure is ready to support it.
An AI infrastructure review can help customers assess network capacity, visibility, security, data governance, and operational risk. It also opens the door to broader conversations around long-term readiness, resilience, and growth.
Equip customer-facing teams to sell business outcomes
Customers rarely invest in technology for technology's sake. They invest to solve a business challenge, reduce risk, or create new opportunities.
Partners should ensure both sales and technical teams can clearly articulate business value. As Roy puts it, every conversation should connect back to one of three outcomes: helping customers make money, save money, or reduce risk. Whether engaging with a CIO, CFO, CTO, or CISO, the ability to link technology to outcomes is what moves the conversation forward.
Build a differentiated, industry-specific services proposition
As customer environments become more complex, the opportunity extends beyond individual products. Organisations are increasingly looking for partners that can bring together technologies, manage multi-vendor ecosystems, and provide trusted guidance.
Developing expertise in key verticals, combined with capabilities in orchestration, visibility, and managed services, can help partners stand out in a crowded market and strengthen their role as trusted advisers.
The conversation: Roy Verboeket on AI, infrastructure, and value
The following transcript has been lightly edited for clarity.
Viktoria Szindekovics: What's the technology trend everyone is talking about, and the one they should be paying attention to instead?
Roy Verboeket: I'm going to be kicking in open doors here, but everybody's talking about AI and thinking about which model is smartest, which tools write the best emails. That's pretty much the headline. But I think what people should be paying attention to, on an enterprise level, on a business level, is the network that has to support the AI transition. Here's the reality. A network built five years ago was designed around one assumption: humans clicking buttons. Somebody opens a browser, does a web search, a server responds. Predictable, manageable, human-speed. AI changes that. AI is machine-generated, never sleeps, runs 24/7, continuously. Every time a user runs an AI query, dozens of backend transactions fire off automatically—databases getting searched, models getting called, results getting assembled—and then multiply that across the number of employees doing that all day long. The way I position it is: it's like a road perfectly designed for family cars. Overnight, container trucks start running on it, 24 hours a day. The road itself isn't broken. The whole system just wasn't built for that. So the trend everyone is talking about is AI, but the one they should be looking out for is what this is going to do to the network.
Viktoria Szindekovics: Very true. I think there were some tests you could hear in the news as well that they started going rogue, some of them. So that's why you need to, as you said, make sure that it's secure, compliant. There's rules set. Partners often get caught up in technology features.
Roy Verboeket: Yes, 100%. I mean, these challenges are showing up more and we see it, we hear it in conversations. I mean, I think you guys on Westcon side, you probably have the same experience. A couple of years ago, I think it was like November 22, ChatGPT came to life. And nobody's really thinking about it. And I think now we're living in 2026. It's August ‘26. We're seeing more and more customers realising that this whole idea of AI is not just about making your workforce smarter. So instead of working harder, you work smarter. They also start to realise it comes with compliancy, gaps and risks. It comes with security validations, ethical ways of doing things. I mean, you just called it out, Viktoria, which is the, you know, the models going rogue. I mean, let's be fair on this one. I don't believe OpenAI or Claude or any of the other ones that were in the news recently.
Roy Verboeket: They didn't do this because they wanted to do something malicious. It just happened. And it shows you how AI has evolved over the past couple of years, but also the speed at which it's still evolving. We don't know what we don't know yet. But the foundational element is that road network built for family cars, now all of a sudden transporting the biggest trucks—and the trucks are getting bigger, faster, heavier. Customers will look back and say: we didn't know what we didn't know. Now we do. How do we deal with it? So that's how I try to pave the road a little bit, by just having that conversation. As long as people are aware, nobody can come back and say you didn't tell them.
Dave Garwood: Yeah, that's a really good way of visualising it—the roads, the expansion, the speed and the size. And I guess if you're a partner, it's not just about selling the AI story, but also making sure that the end-user has all the infrastructure in place: the governance, the automation, the security, to be comfortable that the deployment is ultimately capable of delivering on what it's supposed to be doing without creating problems for the business.
Roy Verboeket: Yeah, and that's a really good point, Dave. The natural conversations we've had over the past couple of years are all about being that trusted advisor to the customer. Here's the thing—there's no badge that gives you that designation. You have to earn it. And the way to earn it is by being honest and truthful with customers, calling out the risk. You can have the biggest and brightest minds in your company with the best models out there. But if you don't have visibility, or you don't have governance in place, or you forget about ethical AI—just making sure everything stays in check and under your control—then you can't manage that risk while allowing your company to grow. Sometimes that means accepting a growth trajectory that's a little bit slower but done right. Rather than the ones that skyrocket and then, three months later, uh-oh.
Viktoria Szindekovics: Yes, planning and being transparent—I think that's crucial.
Roy Verboeket: Absolutely.
Viktoria Szindekovics: Partners often get caught up in technology features. When does a technology conversation become a business conversation?
Roy Verboeket: The easy answer is: the moment you stop talking about what something does and start talking about what it costs when it doesn't. People always look at capabilities. But when those capabilities stop functioning, what's the impact? Take a retailer with self-scan devices in a grocery store. If something happens to the Wi-Fi, depending on the store and the volume of business they do hourly, you're in trouble. Telling them the Wi-Fi is faster is a tech conversation. Telling them that every hour their store network is down costs them this much in lost sales—that's a business conversation. I have a framework I've used for years: three questions—do I make them money, do I save them money, do I reduce their risk? If you can answer yes to any one of those, you have a business conversation. You just need to find that angle and go deeper. And who's sitting across the table makes a difference. A network admin has different priorities from a CFO, CTO, CIO, or CISO. You need to adjust.
Dave Garwood: It's a challenge for some people who've grown up as engineers to make that switch. But to be a really good sales engineer, you need to be able to jump between the two—understand the technology but deliver it in a way the customer understands.
Roy Verboeket: That's it. And that's why we still must make that shift from systems engineering to sales engineering. You need both. But being a sales engineer means looking at technology in a different light. There's still work to be done, but I think it'll flush out.
Viktoria Szindekovics: What's the biggest mistake organisations make when trying to turn emerging technologies like AI into real outcomes?
Roy Verboeket: Companies treat AI like something they buy off the shelf and expect magic. They don't realise the underlying foundation is data, and if the data is a mess, it doesn't matter how good the tools are. I presented this to a customer recently. In the small talk before the meeting I found out he was a huge fan of cooking—and I could see he had about five ovens in his background. So, I used that. Buying a top-of-the-line oven doesn't make you a great cook overnight. Without good ingredients, a fully functional kitchen, and someone who knows what they're doing, all you'll get is an overcooked meal. AI is the same. It's a tool. And without the right structure around it, you'll waste money without knowing what you're attacking. The other piece is guardrails. You don't want your data going rogue and ending up at a competitor because you forgot to put governance in place. And the CISO is becoming a more important conversation partner. In healthcare, for example, there are very strict guidelines around patient data—things like PII. Even if you're selling a switch and an AP, your infrastructure has some visibility into things like IP addresses and MAC addresses, so the governance needs to be there.
Viktoria Szindekovics: If you were leading a partner business today, where would you be investing to stay relevant over the next five years?
Roy Verboeket: Three things. First, train people to talk about business value, not just specs. We've covered that. Second, the
Roy Verboeket: Three areas. The first is business value training. Enable your teams to have financial and risk-based conversations with C-level buyers, the CFO, the CISO, not just the IT manager. That's a different language, and your people need to speak it fluently.
The second is managing the whole system, not just one box. Manage the entire technology kitchen, not just the oven. Customers have multiple vendor components, and someone needs to orchestrate all of it. If you can be the one who does that, you're indispensable.
The third is vertical expertise. Invest in hiring or training people who speak the specific language of an industry, whether that's healthcare or retail. When you can talk to a hospital about their world, in their words, you stop being a supplier and start being a partner.
Viktoria Szindekovics: It can be quite technical, like Platform One, fabric. But if you list the benefits and make it simple, that's what lands.
Roy Verboeket: Right. Simple wins every time.
Viktoria Szindekovics: What technology capability do customers consistently undervalue until they experience the impact firsthand?
Roy Verboeket: Visibility. Every time. It's like the smoke detectors in your house. You buy the fancy cloud-connected one, and you complain about the cost. Why am I paying this much for a smoke detector? You complain right up until the day it beeps and you're able to put out the fire on the stove before it takes the whole house.
Let me give you a real example. There was a school in the Netherlands where the network crashed every single morning, right around nine o'clock. Nobody could work out why. It took days to diagnose. What was happening was that thousands of student devices, iPads, smartwatches, phones, were all hitting the same old access points at exactly the same time, every morning, as the kids arrived.
Proper network visibility would have shown that trend building long before it caused a total crash. You'd have seen the pattern, moved the load, fixed it quietly. Instead, they were firefighting every morning without knowing what they were fighting. That's what visibility buys you: you see the smoke before the fire.
Viktoria Szindekovics: Especially today's demographic. They want everything done instantly. If the system is down, they get frustrated. And the quicker we act, the better the outcome is.
Roy Verboeket: Exactly. And visibility is what lets you act quickly, or better still, before anyone even notices.
Viktoria Szindekovics: Looking ahead, what will separate partners that thrive from those that simply keep up?
Roy Verboeket: Keeping up means you're sitting there waiting to be told what to sell next. You react. Someone releases a product, you learn it, you sell it, you wait for the next one. The ones that actually take a giant step forward already know what's coming. They invest in AI knowledge early, they anticipate the shift, and they act as trusted advisors.
It's like a waiter that takes your order versus a waiter that knows exactly the wine list. The first one writes down what you say and walks off. The second one understands the entire menu, knows which wine goes with which dish, guides your choices, and earns your absolute trust. You'll go back to that restaurant, and you'll ask for that waiter by name.
That's the difference. Keeping up keeps you in the game. Thriving means the customer comes to you first, because you already know the answer before they've asked the question.
