Extreme demos AI-based network assistant

Extreme Networks is expanding its AI capabilities with new technology for its cloud network management platform that’s designed to give customers detailed responses to queries about network anomalies, configurations and documentation to help them proactively troubleshoot and address network issues.

At its Extreme Connect customer enclave this week, the vendor demonstrated Extreme AI Expert. This generative AI-based offering will let customers ask natural language questions about network issues, such as “show me all network anomalies,” and it will generate specific charts and other details about the customer’s network, what anomalies might be causing problems, and what the customer can do to address them, Nabil Bukhari, chief product and technology officer at Extreme Networks, told the audience.

“We have 700,000 pages of documentation on our Web site, so no matter how amazing your search capabilities are, it’s typically going to take you a little time to sift through that data to get answers,” Bukhari said. “We are simplifying that task by offering a system where you can ask a question and the system will come back with a tailored response.”

For example, a company might have forgotten how to configure a secure Wi Fi network, or have a new team member that it wants to onboard, Bukhari said. “So instead of searching thousands of pages, you just ask the system, ‘so how do I configure secure Wi-Fi network?’ The system will tailor a response unique to your question,” he said. In addition to validating the response, Extreme AI Expert will also expose the user to the information used to generate the response. “We make sure that we have the latest documentation integrated, and it gives you a reference that you can just follow up and make sure. That trust is important, [so] that you know where we generated,” Bukhari said. 

The AI Expert gathers data from a combination of Extreme’s public repository, knowledge base and Global Technical Assistance Center (GTAC) documentation, as well as customer network details. It pulls and combines data from applications and devices across the network to establish intelligence related to performance and experience, Extreme stated. 

“Extreme AI Expert will curate enterprise data to provide insights, automate operations and create alerts when it detects anomalies like network overload, degradation or Wi-Fi dead spots, among others,” Extreme stated. It can recommend preventative actions and suggest network optimization based on business KPIs.

AI Expert is currently a tech preview, and Extreme expects to start integrating the technology into its product portfolio – likely in its ExtremeCloud IQ wireless and wired network-management offering, as well as its CoPilot AI-based management tool.

Network vendors ready AI assistants

Extreme joins a number of vendors, including Cisco, Fortinet and HPE Aruba, that have recently unveiled AI-based assistants to help enterprise customers better assess security situations, eliminate errors and automate complex tasks.

• Cisco, for example, added an AI Assistant for Security that will be implemented as part of the vendor’s cloud-based Firewall Management Center and Cisco Defense Orchestrator services. Cisco’s Firewall Management Center is a centralized platform for configuring, monitoring, troubleshooting and controlling Cisco Firepower Next-Generation Firewalls. The orchestrator platform lets customers centrally manage, control and automate security policies across multiple cloud-native security systems.
• Fortinet has is building an AI-based assistant to help customers fight off cybersecurity threats. Fortinet Advisor is aimed at helping security operations teams make more informed decisions, respond to threats faster, and simplify routine and complex tasks. The AI-based assistant will be available as part of Fortinet’s FortiSIEM security information and event management platform and FortiSOAR security orchestration, automation, and response product.
• HPE Aruba is deploying genAI-driven search tools in its management platform to help customers access detailed responses to queries about network configurations, documentation, and other IT operational issues. The company is incorporating multiple HPE-trained large language models (LLM) into the search tools that are part of HPE Aruba Networking Central, its core cloud-based management and orchestration platform for wired and wireless networks spanning campus, branch and data center sites.

AI tools bolster IT operations

Generative AI functionality is something a lot of organizations are curious about, according to IDC research manager Brandon Butler, who said that these types of assistants and automation tools represent an easy onramp into the world of AIOps.

“A common theme we’ve seen from these GenAI-inspired systems is to make it easier for customers to find information about the company’s products … and combine that with data specific to a customer’s environment,” Butler said. “A key to vendors doing this successfully is the quality of the responses an AIOps system provides, and ensuring that customer data is handled securely by the vendor.”

If the last part is handled correctly, he added, AIOps systems can provide meaningful assistance on a wide range of tasks, from optimizing network performance to diagnosing and remediating network issues.

Butler’s colleague, IDC research director Mark Leary, concurred and noted that there are two key strengths necessary to the success of the type of AI assistant feature rolled out by Extreme. First, he said, the data contained within the LLMs contributing to the assistant must be complete, extending well beyond basic documentation. “The broader and more real the data feeding the supporting LLM, the better off the resulting guidance,” Leary said.

Second, responses need to be tailored to the particular operators making specific inquiries, taking into account network configuration, environment and the operator’s requirements. “No two networks look alike, so the ability to provide customized responses is a key differentiator for both these ‘assistants,’ and, ultimately, for the inquiring operator,” said Leary.

Senior writer Jon Gold contributed to this story.