{
  "id": 17279605,
  "title": "Extreme Networks wants the network to run AI, not just be run by it",
  "url": "https://urgent.news/2026/10/30/extreme-networks-wants-the-network-to-run-ai-not-just-be-run-by-it",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-30T19:53:09.000Z",
  "source": {
    "name": "SiliconANGLE",
    "slug": "siliconangle",
    "url": "https://siliconangle.com/2026/10/30/extreme-networks-wants-the-network-to-run-ai-not-just-be-run-by-it/"
  },
  "original_language": "en",
  "account": "At Extreme Networks' recent AI Summit in Amsterdam, the company unveiled Agent ONE Coworker and introduced the concept of Exchange zones, showcasing how access points and switch compute can be repurposed as on-premises AI infrastructure. This shift in perspective suggests that AI should manage the network and serve as a platform for AI applications, rather than merely being managed by it.\n\nDuring the summit, Extreme's chief technology officer and president of AI platforms, Nabil Bukhari, argued that the first wave of AI in networking fell short because it relied on a chatbot built on fragmented data. This design flaw made it difficult for administrators to engage with powerful AI assistants, as they often asked it questions that were already answered on the dashboard.\n\nTo address this issue, Extreme introduced Agent ONE Coworker, an ambient AI agent built into Extreme Platform ONE. This agent has demonstrated significant results, with customers saving about 1,000 hours of troubleshooting, 41% of Platform ONE customers using it for daily tasks, and users responding to 98% of critical alerts.\n\nMoreover, Extreme is set to launch Exchange zones in the first half of 2027. These zones combine Extreme Universal Compute Platforms or customer-owned servers with spare capacity from existing access points and switches at a site. AI applications from the Extreme Exchange marketplace will run locally on open-weight models, with cloud inference used only when policy allows, and all actions logged for transparency.\n\nBukhari emphasized that AI agents like Agent ONE Coworker should prompt administrators instead of the other way around. This change in approach allows small teams to achieve the productivity of larger teams. However, Bukhari acknowledged that there is still room for improvement, as 58% of support cases opened by Agent ONE arrive with all required evidence.\n\nExtreme has also implemented guardrails to ensure that Agent ONE Operator mode operates safely. Each of the 268 tools can be configured to auto-authorize, ask first, or never use, with four baseline rules governing its actions. These rules include not harming the network, never exposing credentials, never acting outside its role, and never acting in another tenant.\n\nThe strategic value of Extreme's approach lies in the rich, real-time telemetry generated by networks, which currently goes unused in most cases. By repurposing this data for AI applications, organizations can unlock opportunities in areas like sovereignty, localization, survivability, and economics. Furthermore, the compute required for these applications is often underutilized within access points, providing an opportunity to maximize existing investments.",
  "summary": "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 […] The post Extreme Networks wants the…",
  "key_points": [
    "Extreme Networks unveils Agent ONE Coworker at AI Summit in Amsterdam",
    "Agent ONE Coworker saves 1,000 hours of troubleshooting and improves daily task usage by 41%"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}