Juniper Networks, one of the early adopters of AI networking, is evolving its Mist AI platform, with a goal of unifying AI tools for wired, wireless and WANs under one suite.
Historically, the Mist brand was associated with Wi-Fi, and the company recently expanded it to the campus. Now, it is the single platform for AI across a Juniper environment.
Parts of the business — such as data center, which historically fell under the Apstra brand — will also shift focus to align with Mist. The company also announced that its Mist AI engine will now be Marvis AI.
To clarify, Juniper Mist is a broader, AI-powered networking platform, while Marvis is the conversational AI assistant within Mist. Mist is the AI engine and cloud-based platform, and Marvis is the UI that provides real-time answers and guidance to IT staff. This can be confusing, and Juniper is aiming to bring better clarity with this launch.
Some of the latest updates focus on the following:
-
Marvis Minis, which are network digital twins.
-
Marvis Actions, which provide a proactive method of resolving user issues.
-
Marvis Client, which is the software agent.
Prior to the launch, Christian Gilby, senior director of product marketing at Juniper, told me the company is focusing on Marvis with this launch.
“What we’re introducing here is more visibility into the self-driving action,” Gilby said. “The challenge is that some IT teams are ready to go there, while others don’t have the confidence yet.”
Marvis Minis Adds End-to-End Monitoring
Marvis Minis, which already provides full-stack monitoring across wired, wireless and WAN networks, now includes what Juniper calls “client-to-cloud” monitoring. This enables IT teams to track the entire user experience and identify performance issues more precisely, without deploying extra hardware or software.
Juniper rolled out two new service-level metrics. One measures network performance across the WAN, and the other tracks how well applications are running. The biggest improvement is the ability to monitor app performance from the WAN to the cloud. This means IT teams can test and confirm that apps are working properly before users even notice a problem.
Dashboard Updates Improve Automation Oversight
Juniper also updated its Marvis Actions dashboard to give IT teams better oversight of automated network operations. The new interface shows which actions are handled automatically and which require manual approval. Gilby said IT teams can choose how much autonomy to allow and can authorize or restrict specific self-driving actions.
“You don’t want the IT team to have to go and hunt for problems and solve them,” Gilby said. “So, Marvis Actions is always analyzing the data, finding the problems, and then either automatically resolving them or sending alerts.”
As part of this update, Marvis can now identify and fix misconfigured virtual LAN tags on its own, if authorized to do so. The dashboard also keeps a log of all actions taken, giving teams visibility into how issues were resolved.
Currently, Marvis’s self-driving features only work on Juniper’s hardware and systems. For example, it can automatically fix configuration issues on Juniper switches or routers, but not third-party products. In the future, Juniper wants to be able to authorize Marvis as an agent on third-party systems, so it can make automated changes across a wider range of network equipment, not just Juniper’s, Gilby said.
Marvis Client Extends Monitoring to User Devices
In addition to monitoring the network side, the new Marvis Client focuses on collecting data directly from user devices. This helps detect issues that might not be visible from the access point (AP). For instance, if a device has trouble with the AP, the network might not catch the problem unless it can also see the device’s data.
Marvis Client is a software agent that runs on Android, Windows (desktop and mobile) and macOS devices. It pulls telemetry directly from the user’s device, offering a clearer picture of what’s happening. Therefore, IT teams can spot patterns, such as whether a specific firmware version is causing issues, and make informed decisions.
Juniper’s approach offers an alternative to hardware-based monitoring options, which require physical sensors. This is especially valuable in industries where devices are spread out and reliable connectivity is critical.
“The big driver for customers has been in the warehouse retail space, because there are a lot of IoT devices,” Gilby said. “They just don’t have enough people in different areas, and it’s a challenging environment with all the metal. The other one is healthcare. In a hospital, you can’t afford for a medical device not to be able to connect.”
Focus on Visibility and Gradual Automation
These updates reflect a broader shift in how Juniper is positioning its AI capabilities, bringing them together under one platform. Juniper is giving IT teams more visibility into network operations and building confidence in automated processes. By showing what actions Marvis takes and how issues are addressed, Juniper is allowing customers to automate at their own pace.
