Extreme says AI coworker saved 1,000 hours in month

Customers using Extreme Networks' Agent ONE Coworker tool saved an estimated 1,000 hours of troubleshooting in its first month of availability, according to the company. Users also responded to 98% of the system's critical nudges.

The figures are part of the first public data on uptake for the artificial intelligence tool. Built into Extreme Platform ONE, it is designed to flag network issues without waiting for users to enter a prompt.

Early engagement with Agent ONE Coworker was about 10 times higher than with Extreme's earlier conversational AI agent, the company said. It added that 41% of Extreme Platform ONE customers used Agent ONE for daily tasks during the first month after launch.

The networking supplier is positioning the product around a more proactive model of AI use in IT operations. Instead of relying on technicians to ask questions through a chatbot-style interface, the system sends nudges intended to highlight potential problems and recommend actions within existing workflows.

Extreme said 58% of support cases opened through Agent ONE Coworker included the evidence support teams needed to begin working on the issue. The company also said the tool helped customers resolve some issues in minutes, compared with what it described as an industry average of two days.

Customer use

One early user cited by the company was Fastcom, a New Zealand IT and telecommunications business and Extreme partner.

"Extreme Agent ONE Coworker enables us to scale our sites and services without the usual configuration sprawl. I've spent years flicking between consoles to chase down one issue, so having networking, security, and AI together in Extreme Platform ONE feels like someone built the tool I would have designed myself," said Brent Addis, Group Security Architect at Fastcom.

Hill College in Texas also described cases in which the system helped identify faults and narrow down security issues on its network.

"Extreme Platform ONE is like having two or three more network engineers on staff, and for a team our size, that changes everything. When HR lost wireless, it could have significantly impacted campus operations, since our admin building also houses Finance, Registration and Enrolment. Agent ONE Coworker told me exactly where to look: one access point needed attention, and several others were due for firmware upgrades. We resolved the issue in about 20 minutes instead of an entire morning and fixed problems we likely would have missed. Additionally, when a device on our guest network turned up running crypto mining software, Agent ONE helped us trace it through our access points and quickly narrow down its location. Extreme Agent ONE Coworker puts expert-level insight in my team's hands the moment we need it," said Kevin Rocha, Chief Information Officer at Hill College.

Touro University in New York was also cited as an early user. Its feedback focused on the value of showing networking teams the underlying data and reasoning behind AI-generated responses.

"AI is only as good as the data it's ingesting. Extreme Agent ONE Coworker separates chaos from insight, giving networking teams greater transparency into the data and reasoning behind its answers," said Moshe Sittner, Wireless Infrastructure Engineer at Touro University.

AI shift

The company's comments reflect a broader shift in enterprise software from assistant-style AI tools to systems that continuously monitor conditions and suggest or trigger actions. In networking and infrastructure management, suppliers are trying to show that these systems can reduce operational workload and shorten the time needed to isolate faults.

Extreme said Agent ONE Coworker runs on an AI stack that combines reasoning, live network intelligence, and operational data. The tool is available to all Platform ONE customers.

Nabil Bukhari, Chief Technology Officer and President of AI Platforms at Extreme, argued that conventional conversational interfaces have limits in environments where users may not know what question to ask first.

"Ultimately, that reduces operational workload so that we can do better work, faster," said Sittner.

"Conversational AI only helps when people already know what to ask. That's a design limitation, not a user problem. We built Extreme Agent ONE Coworker to reason continuously over live network context and act on what it finds, under governance our customers control. Unlike others in our industry who are still focused on dashboards and conversational AI, our customers are already benefiting significantly from the ambient AI in Agent ONE Coworker mode that reasons and acts on your behalf," said Bukhari.