Artificial intelligence is rapidly changing the operational logic of enterprise IT. At Extreme Connect 2026 in Orlando, Extreme Networks presented a strategy that places AI agents at the center of enterprise networking. The company argues that modern networks can no longer rely on manual administration and fragmented tools, particularly as AI workloads, cloud applications, and connected devices continue to increase operational complexity.
Extreme Networks used its annual Extreme Connect conference in Orlando to present a broader vision for enterprise networking built around autonomous AI systems. The company introduced Agent ONE, a framework designed to move enterprise IT operations beyond assistive AI toward continuously operating AI agents capable of monitoring, analyzing, and executing network tasks in real time.
The announcement reflects a wider industry trend in which networking vendors are integrating AI more deeply into infrastructure management. While earlier generations of AI tools focused primarily on providing recommendations or simplifying workflows, Extreme Networks is positioning its new platform as an operational layer capable of autonomous action within governance boundaries defined by IT teams.
At the center of the strategy is what the company calls the “Extreme AI Stack,” a four-layer architecture that combines reasoning models, contextual network intelligence, procedural knowledge, and autonomous execution.
The first layer focuses on AI reasoning infrastructure. Extreme Networks says the platform uses advanced frontier AI models from major cloud and AI providers to continuously improve analytical capabilities as models evolve.
The second layer, AI Core, is designed to provide contextual understanding of enterprise networks. According to the company, the system combines live and historical network telemetry, domain-specific knowledge, and relationship mapping between users, devices, policies, and infrastructure. The objective is to allow AI systems to understand operational context rather than simply process generic prompts.
The third layer, referred to as the Skills Layer, introduces operational procedures and integrations into the AI framework. This includes encoded networking workflows, integrations with identity systems, cloud platforms, security environments, and IT operations tools. Extreme argues that procedural knowledge is critical because enterprise networking depends not only on technical reasoning, but also on organizational governance, maintenance windows, escalation policies, and operational safeguards.
The final layer is the agentic execution framework, where specialized AI agents operate in parallel to validate configurations, assess security posture, troubleshoot faults, and execute tasks. Extreme Networks says the framework includes governance controls, auditability, memory systems, and orchestration functions designed to make autonomous operations suitable for enterprise environments.
A major part of the announcement focused on Agent ONE Coworker, the first operational mode of the platform. Unlike conventional chat-based AI assistants, the system is designed to proactively engage with IT teams. The platform can monitor network activity, identify anomalies, investigate incidents, and deliver recommendations or automated actions without waiting for manual prompts.
Extreme Networks introduced a feature called “Nudge,” which allows the system to communicate context-sensitive alerts and recommendations. Rather than generating constant notifications, the company says the platform is intended to adapt communication based on urgency and operational relevance.
The company also announced Agent ONE Operator, a second operational mode expected later in 2026. While Coworker focuses on collaborative assistance, Operator is positioned as an always-on autonomous operational engine. The system is intended to execute recurring operational tasks, respond to incidents independently, and maintain continuous oversight of enterprise infrastructure.
Extreme Networks described the Operator model as a shift from reactive administration toward persistent operational automation. The company says the platform will support event-triggered tasks, scheduled processes, and continuously running monitoring functions.
To support this approach, Extreme Networks introduced a memory architecture modeled on different forms of human memory, including working memory, episodic memory, semantic memory, and procedural memory. According to the company, these systems are intended to help AI agents adapt to organizational practices and operational history over time.
The broader platform strategy also includes the expansion of Extreme Platform ONE, which now supports third-party device management, integrated zero trust security workflows, certificate lifecycle management, and simplified enterprise licensing models.
In parallel with the AI announcements, Extreme Networks expanded its Wi‑Fi 7 portfolio with new indoor and outdoor access points designed for high-density venues, healthcare, education, industrial environments, and hospitality deployments. The company says demand for AI workloads, IoT systems, and low-latency applications is accelerating the need for higher-capacity wireless infrastructure.
The announcements arrive as Extreme Networks reports continued financial growth. For the third quarter of fiscal 2026, the company reported revenue growth of 11 percent year over year and SaaS annual recurring revenue growth of nearly 29 percent. The company also highlighted new customer deployments across healthcare, education, aviation, retail, and sports venues.
The strategy presented at Extreme Connect 2026 signals how networking vendors are redefining enterprise infrastructure management around autonomous AI operations. Rather than positioning AI as a separate productivity tool, Extreme Networks is attempting to integrate AI directly into the operational core of enterprise networking.
Whether enterprises fully adopt autonomous networking models will likely depend on governance, trust, security transparency, and operational reliability. However, the direction presented by Extreme Networks suggests that AI-driven infrastructure management is moving from experimental deployments toward operational integration.
Dr. Jakob Jung is Editor-in-Chief of Security Storage and Channel Germany. He has been working in IT journalism for more than 20 years. His career includes Computer Reseller News, Heise Resale, Informationweek, Techtarget (storage and data center) and ChannelBiz. He also freelances for numerous IT publications, including Computerwoche, Channelpartner, IT-Business, Storage-Insider and ZDnet. His main topics are channel, storage, security, data center, ERP and CRM.
Contact via Mail: jakob.jung@security-storage-und-channel-germany.de