Extreme Networks：讓網絡不只由 AI 管理，也運行 AI

At its recent AI Summit in Amsterdam, Extreme Networks Inc. shared early Agent ONE Coworker results and previewed Exchange zones, which put spare access points and switch compute to work as on-premises artificial intelligence infrastructure. Here’s what it means and what information technology pros should do about it.

Most of the conversation about AI and networking has flowed in one direction. AI comes in, examines the network and helps the network team run it better. That’s useful, but it treats the network as the patient, not as a resource.

At the summit, the company argued that there is a better way. The network should be managed by AI and serve as a place where AI runs, using data that most organizations currently throw away. This would let network teams stop being the people who merely keep the Wi-Fi up for the AI project and become the people who host it.

The news

Extreme made two announcements. The first was a progress report on Extreme Agent ONE Coworker, the ambient AI agent built into Extreme Platform ONE. In its first month, Extreme says, customers saved about 1,000 hours of troubleshooting, 41% of Platform ONE customers used it for daily tasks, and users responded to 98% of critical “nudges,” the proactive alerts the agent raises after running its own pre-investigation. Engagement was roughly 10 times that of Extreme’s first-generation conversational assistant.

The second was a preview of Extreme Exchange zones, planned for the first half of 2027. A zone combines an Extreme Universal Compute Platform or a customer-owned server with spare capacity from Extreme access points and switches already installed at a site. AI applications from the Extreme Exchange marketplace run locally on open-weight models by default, with cloud inference used only when policy allows and logged every time. Agent ONE Operator mode, the autonomous tier, also arrives in the first half of 2027.

Ambient beats conversational

The first wave of AI in networking wasn’t a smashing success because it relied on a chatbot built on fragmented data. During his talk, Nabil Bukhari (pictured), Extreme’s chief technology officer and president of AI platforms, called it a design flaw. “Conversational AI requires the user to have the question and the context, and you will be surprised how difficult that is when it comes to technology domains,” he said.

In Extreme’s focus groups, many administrators asked a powerful AI assistant how many devices they had, even though the number was already on the dashboard. Coworker flips that. “We have moved to the world where instead of you prompting AI, AI prompts you,” Bukhari said.

The best evidence came from a customer. Kevin Rocha, chief information officer at Hill College, described a wireless outage in a building housing HR, finance and enrollment. “Agent ONE Coworker told me exactly where to look: One access point needed attention, and several more were due for firmware upgrades. We resolved the issue in about 20 minutes rather than an entire morning,” he said. Platform ONE, he added, “is like having two or three more network engineers on staff.”

That’s the value of AI for networking in most enterprises. It doesn’t replace the network team. It gives a small team the reach of a larger one. Bukhari also admitted that there is more work to be done. He explained that 58% of support cases opened by Agent ONE arrive with all the evidence support needs. “Fifty-eight is pretty good, but 58 is still 58 and not 100.”

Governance has to live outside the agent

Andrew Grill, the opening keynote speaker and author of “Digitally Curious,” discussed the stakes. “When a chatbot gets it wrong, you might have a bad paragraph. If an agent gets it wrong, you have a bad Tuesday,” he said. His test for any deployment: “If your agent does something wrong at 2 in the morning, whose phone rings?”

Extreme’s Operator demo is subject to guardrails. Each of the 268 tools can be configured to auto-authorize, ask first or never use. Four baseline rules apply to everything: do no harm to the network, never expose a credential, never act outside your role and never act in another tenant. Every task has a full audit trace.

An important design criterion is that governance reside within Extreme Exchange, not Platform ONE. “You can’t govern something from within,” Bukhari said. “That is how you get rogue agents.” If the system running the agent also controls its permissions, you don’t have a control plane. You have an honor system.

Exchange zones are the bigger bet

Networks generate dense, real-time telemetry on location, movement and behavior that far exceeds what operations needs, and cloud platforms drop most of it because moving it is expensive. Bukhari said the valuable uses tend to hit four walls: sovereignty, localization, survivability and economics.

The innovation is where the compute comes from. Higher-end access points have four to eight cores and typically use only one, so organizations already pay for and power idle compute hanging from their ceilings. In one healthcare demo, a fall-detection app that combined camera feeds, Wi-Fi location and paging to find the closest responder ran entirely on access points. A second demo used Seattle Seahawks game data to predict gate congestion and showed that some fans faced a seven- to eight-minute walk to the nearest pro shop.

“This is not a network management use case,” Bukhari said. “These use cases are impossible without that rich, real-time telemetry.” That’s the strategic point. The return goes to operations, facilities and fan experience, giving network leaders something they’ve rarely had: a revenue story.

Zones will ship in 2027, and the demo apps were built in collaboration with design partners. The model needs a real application catalog, so Extreme is pairing it with a free, invitation-only forward-deployed engineering program. Pricing will be consumption-based, measured in “agent actions.” As Grill noted, most organizations haven’t priced the cost of their tasks today, so they’ll struggle to judge value.

Sovereignty is about permission

As expected, the most valuable insights came from the customer panel. Jonathan Monk, CIO of the Institute of Cancer Research in London, manages roughly 9 petabytes of data, but performance isn’t his primary reason for building sovereign AI. “When we ask people in our organization what’s the biggest thing that holds them back from using AI, it’s that I’m not sure if I’m allowed to,” he said.

Andrew Beynon, chief enterprise architect at Extreme partner ARO, linked that to shadow AI. The goal, he said, is “to make the journey for legitimate adoption of AI easier than the move to illegitimate adoption.” Monk called it building “paved roads.” The industry treats sovereignty as a compliance cost, but these customers see it as an adoption accelerator.

David O’Neill, director of IT operations at Liverpool Football Club, offered a balanced take on AI. “While AI is an answer, it’s not necessarily the answer.” Bukhari highlighted the next challenge: “Adoption is no longer the problem. It’s the optimization of adoption.” After AI-generated product specs ballooned to 200 pages, Extreme capped them at three pages.

Advice for IT pros

Include the network team in the AI plan. If it isn’t involved, decisions about traffic, inference location and data residency will be made without the people who understand them.

Separate governance from execution. Check whether agent permissions, audit logs and the kill switch are outside the system the agent runs in.

Inventory the data your network discards. Identify the location and occupancy telemetry you already generate, and which business units would pay to use them.

Pressure-test the edge-compute math. Ask how workloads are prioritized during RF demand spikes and which model sizes realistically fit on an access point.

Price your tasks first. Without a baseline cost for today’s work, it’s impossible to evaluate consumption-based AI pricing.

Final thoughts

Extreme can’t outspend its larger competitors, so it’s betting on speed and focus as a pure-play networking company. The Agent ONE numbers suggest that’s paying off on the operations side.

Exchange zones are the bolder idea. Turning installed access points and switches into a distributed, centrally governed inference fabric is more ambitious than anything I’ve seen from Extreme’s larger rivals. It’s also unproven, and 2027 is a long way off in AI time. As Beynon put it, “today’s knowledge has got an expiry date.”

The direction Extreme is headed is the right one. For 30 years, the network has been the most important infrastructure, yet nobody wanted to talk about it until it broke. AI makes the network the system that connects data, compute and people, and possibly the one that runs the AI, too. Network pros who see that early will have a much bigger role in where their organizations go next.

Zeus Kerravala is a principal analyst at ZK Research, a division of Kerravala Consulting. He wrote this article for SiliconANGLE.

Photo: Zeus Kerravala

A message from John Furrier, co-founder of SiliconANGLE:

Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities.

• 15M+ viewers of theCUBE videos, powering conversations across AI, cloud, cybersecurity and more
• 11.4k+ theCUBE alumni — Connect with more than 11,400 tech and business leaders shaping the future through a unique trusted-based network

SiliconANGLE Media is a recognized leader in digital media innovation, uniting breakthrough technology, strategic insights and real-time audience engagement. As the parent company of SiliconANGLE, theCUBE Network, theCUBE Research, CUBE365, theCUBE AI and theCUBE SuperStudios — with flagship locations in Silicon Valley and the New York Stock Exchange — SiliconANGLE Media operates at the intersection of media, technology and AI.

Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.