Juniper Mist extended its artificial intelligence-driven networking capabilities to the WAN today alongside the company's virtual assistant Marvis. The introduction of SD-WAN functionality now positions Mist to compete directly against Cisco's Meraki and Hewlett Packard Enterprise's Aruba platform.

However, unlike the competition, Mist isn't reactionary, explained Jeff Aaron, head of marketing at Mist in an interview with SDxCentral.

"The challenge with SD-WAN is still it's very static. It's based upon static policies and intent-based initiatives; it's not adapting in real time; it's not true self-driving; and it's not really looking at the user experience," he said. SD-WAN is "still making decisions based on network conditions and in some respects application conditions."

From the start, Juniper set out to solve this problem by using artificial intelligence (AI) that it acquired when it bought Mist Systems to glean insights from clients connecting over wireless, the LAN, and now the WAN.

According to Aaron, this end-to-end approach enables IT teams to quickly get to the bottom of network disruptions. For example, if a Zoom call were to go down, he said, Mist's AI capabilities would be able to identify if there was an issue with the WiFi, LAN, or the internet connection.

Aaron adds that the rise of remote work in response to the COVID-19 pandemic has resulted in new challenges for enterprises.

"Home is where the WAN is," he said. "That's now the new corporate WAN, and IT help desk calls are going up astronomically, and they're harder than hell to troubleshoot if one person is using a Netgear access point and another is using a DLink."

While the enterprise may not be able to control which service provider the employee uses, giving them the right equipment can go a long way to alleviating these kinds of headaches, said Aaron.

However, visibility across the WAN is just one piece of the puzzle. Alongside Mist's WAN capabilities, the company also rolled out its virtual assistant called Marvis.

Marvis takes Mist's existing AIOps platform and makes it completely interactive. But rather than telling you the weather, playing some tunes, or turning on your lights like Amazon's Alexa might, Marvis is much more interested in helping customers identify anomalies and suggest solutions.

According to Aaron, this is a clear differentiator for Mist. "Vendor's like Aruba and Cisco, their AI is all peer-based," he said. "It's like Walmart is a retail company, so I can tell Walmart what other retail companies are doing, but I can't tell Walmart what's happening in their stores. For us that's ridiculous. Every store is different and every company is different, so if you can't do AI-based intelligence down to the client level and the site level, it almost defeats the purpose."

"Marvis is a full conversational interface," Aaron explained. Users interact with the virtual assistant much like they would a chat bot.

For example, a customer experiencing problems with their network could simply type "what's wrong with my network," and Marvis would chew through the data, identify the anomaly and suggest a resolution to the problem.

This kind of interactivity will make it easier for IT teams to address challenges before employees even notice there is a problem, Aaron said.

"We're literally collecting over 150 user states every two seconds from every access point, for every client device and similarly we're pulling in data from EX switches, telemetry data, SRX series firewalls for SD-WAN, and that's all going to the cloud and being munged by our AI engine," he said.

This also allows Mist to track and maintain experiences at a per-user basis.