This article is part of an AI Research Insight Series paid for by Juniper Networks.

“Experience is the new uptime,” says Jeff Aaron, VP of enterprise marketing at Juniper Networks. “It’s not enough to look at your dashboard and say, ‘My network is running all green.’ That doesn’t mean your users are having a good experience.”

The old network-centric way of operating networks simply can’t keep pace with the soaring number of applications, users and devices within an organization. So it’s no surprise that AI is now powering experience-first networking for both users and operators – and in the process, reducing costs, improving efficiencies and boosting end-user satisfaction.

This is true whether you’re a retail company maintaining its brand experience with consumers, a school ensuring their students and teachers are happy, or an enterprise making sure employees working in a hybrid work environment, at home or onsite, have predictable, reliable, measurable services that make them more productive.

So, what makes AI-driven operations, or AIOps, essential for IT leaders? And how does it power an experience-first approach?

The challenge

An AI-powered strategy is more important than ever as the demands on IT departments keep growing. Outside of the pandemic, there’s the astronomical growth in devices, apps and operating systems. In the throes of the pandemic, there’s the urgent need to offer users the ability to work anywhere.

“Now everyone’s home has become a branch office,” Aaron says. “How do you deliver reliable and predictable networking and security to all these locations?” And that will remain an issue even after workers return to the office since the pandemic has entrenched ongoing work-from-anywhere situations.

The second big challenge is bringing workers back safely. IT now needs to be working hand-in-hand with facilities to embed pandemic planning as part of disaster recovery. There are practical considerations — do you maintain an open seating environment, or keep people distanced? Are workers invited back in droves, or should you move into a hoteling environment, where people can reserve seats when they show up?

“All that impacts your network, your wireless and things like location services,” he says. “And preparing for, worst case scenario, if another similar situation requires contact tracing, social distancing, alerting employees and more. That’s the new norm. It has to be planned for.”

Where AI comes in

Given these challenges, IT department heads are realizing they need to focus more on end-to-end user experiences. It’s not just about the network anymore, but about the users and what kind of experiences they have on that network.

AIOps makes maintenance and monitoring more automated, provides better insight into user experiences and greater ability to take proactive action.

With automation, insight into your user experiences is continually visible. When you’re able to collect 150 user states from every mobile client, every two seconds, you know exactly what’s going on with every user. And as that data is fed back into the platform, you can constantly optimize. You’re in a prime position to take action as anomalies or patterns are detected, then provide recommendations to operators, so they can proactively get ahead of things before users even know there’s an issue.

AIOps is also an invaluable tool for user support. Ideally, you want to solve a problem before the user has to call or create a ticket, of course. But if they do have to call, an AI solution can help detect and resolve an issue faster to minimize the IT resources required.

Consider the AIOps conversational interface Marvis, Juniper’s virtual network assistant driven by Mist AI. Rather than combing through a thousand event logs across a thousand systems to determine a solution for a user’s problem, an IT team using Marvis can use a simple query like, ‘What was wrong with Jeff’s connection yesterday?,’ and return an actionable answer – ‘It was a wireless issue, and here’s how to fix it.’

The AI assistant can deduce the intent behind questions and provide actionable results based on its knowledge base, with interactive queries to drill down further based on suggested next steps. And the assistant continues to learn based on feedback from users. Built with open APIs, teams can integrate the platform with their current IT systems and automate the network.

“You can argue that AIOps and virtual network assistants are minimizing the need for dashboards, the same way dashboards did for CLI,” says Aaron. “That’s where the industry is moving toward.”

Solutions in the real world

Juniper Networks was the first to recognize that the IT service model has to change, Aaron says. Mist AI brings insights and Self-Driving Network™ automation to optimize end-to-end user experiences and minimize IT costs across LAN, WLAN and WAN.

“The old reactive way of looking at the network needed to shift to one that’s more proactive, automated and focused on the user,” he says. “We designed our solutions from the ground up to do that.”

He points to the case of ServiceNow, a SaaS company that wanted to eliminate all inbound user-generated trouble tickets. In particular, they wanted to be able to detect and resolve a Wi-Fi problem before a user knew it existed, since they were getting over 100 complaints per month.

“Once they put Juniper in place with AI and AIOps, they eliminated over 90% of user-generated trouble tickets. They’re now down to one or two per month and that’s huge,” he says. “That shows you the ability of proactive automation and insight and action when properly applied.”

And Dartmouth College is using Juniper Networks for more reliable wireless access wherever their students roam, with a help desk helmed by grad students who aren’t especially tech savvy. By being able to give their student workers a virtual network assistant that can answer natural language questions and return solutions, the Dartmouth College IT department can focus on more strategic things than just answering simple calls.

The Mist AI advantage

Mist delivered the first AI-driven network almost five years ago, born out of the need to combine network automation with insights for great user experiences.

“Internally we often say that Mist AI is the rocket fuel that raises the bar on what’s possible across all aspects of the AI-driven enterprise,” Aaron says, “and we are continuing to expand the entire Juniper portfolio and bring it all under a common Mist AI umbrella.”

To shift the focus from network and application behavior to actual user experiences, they recently introduced Juniper Mist WAN Assurance to bring AI-driven insights for the WAN and branch to Juniper’s SD-WAN solution.

This cloud-based service streams key telemetry data from Juniper SRX devices to the Mist AI engine in the cloud to enable customizable WAN service levels that are used to optimize user experiences. It’s able to correlate events across the LAN, WLAN and WAN for rapid fault isolation and resolution across all domains, and it enables proactive actions as a result of anomaly detection with automated workflows, with an eye toward completely self-driving networks from client to cloud.

This, along with the Juniper Mist Virtual Network Assistant and the Marvis conversational interface, helps IT administrators to more nimbly address network, application and user issues; corrective actions are recommended or taken automatically to resolve the issue in real time, often before the user is even aware of a problem.

“Our goal is to have something like 80% of all trouble tickets answered by our virtual network assistant,” Aaron says. “We’re pretty close to that. It’s just learning and getting better over time.”

“There will be a day where most trouble tickets can be handled by AI,” he adds. “Mist AI is what provides the automation, the insight, the actions across our wired and wireless and SD-WAN domains.”

Dig deeper: Learn more about experience-first networking and see AIOps in action

Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. Content produced by our editorial team is never influenced by advertisers or sponsors in any way. For more information, contact