Extreme Networks unveiled Agent ONE at its annual Connect 2026 conference in Orlando, introducing what it describes as a second-generation AI agent built specifically for enterprise network operations. The platform moves beyond prompt-based AI assistants to proactive, context-aware agents that can detect network anomalies and act on them autonomously within governance boundaries, according to Computer Weekly.

Agent ONE ships in two phases. Agent ONE Coworker, launching in July 2026, is an interactive agent that continuously monitors network activity, investigates anomalies, and executes changes with human approval. Agent ONE Operator, arriving in Q4 2026, will run workflows autonomously within defined guardrails, responding to events in real time without requiring constant human input, according to Network World.

Four-Layer AI Stack

The platform runs on a purpose-built AI architecture with four layers, as described by Nabil Bukhari, Extreme’s CTO and president of AI platforms, during the keynote. The infrastructure layer provides reasoning via frontier models sourced from hyperscalers. An Extreme AI Core layer encodes a networking-specific knowledge graph covering MACs, clients, policies, sites, and service relationships. A skills layer houses connectors, data pipelines, and workflows that map to customer-specific standard operating procedures. The agentic layer on top is where Agent ONE Coworker and Operator execute, with a governance harness enforcing business rules, according to SiliconANGLE.

“Agent ONE does not wait to be asked. It notices. It reaches out,” Bukhari wrote in a blog post accompanying the launch. “A [severe alert] fires at 2am. Before Agent ONE reaches out it has already investigated. It does not wake you up with an alert. It wakes you up with findings.”

The “Nudge” Capability

A central feature is what Extreme calls “nudge,” delivering contextual recommendations that turn network insights into immediate action. Use cases cited by Bukhari include detecting rising Wi-Fi congestion in a school and recommending or automatically applying a fix, and identifying recurring point-of-sale slowdowns in retail to suggest traffic prioritization during peak hours, according to Network World.

Ron Westfall, vice president and practice lead at Hyperframe Research, called the nudge capability “a psychological pivot in network management,” writing in a LinkedIn post that “the AI is transformed from a passive database into a proactive coworker that identifies invisible issues, such as retail POS lag, before they manifest as critical failures,” as cited by Network World.

Customer Readiness Data

Extreme cited internal customer surveys showing 93% trust AI-powered networking, 94% are ready to invest in networking AI technologies, and 57% expect results from AI products within weeks, according to Computer Weekly. The company also announced Extreme Exchange, an AI skills marketplace for Platform ONE where customers can discover, activate, and manage skills that extend Agent ONE Operator’s capabilities, with support for first-party, partner-developed, and eventually customer-created skills.

Vertical Agent Specialization

The launch reflects a broader pattern in the enterprise agent market: vendors are moving from generic AI assistants toward domain-specific agents tailored to particular infrastructure verticals. Zeus Kerravala, founder of ZK Research, assessed the move by noting that “most vendors are still delivering AI as copilots. Extreme is taking a different path, embedding reasoning, context and execution into the network itself,” according to Computer Weekly. As cloud providers like AWS, Google, and Microsoft build horizontal agent platforms, Extreme is betting that networking infrastructure requires agents with specialized domain knowledge that generic platforms cannot replicate.