In the Juniper Mist dashboard, the Marvis Actions page serves as a centralized "proactive inbox" that identifies critical issues across the organization. However, the depth of information and the ability to interact with these findings are strictly tied to the Marvis Virtual Network Assistant (VNA) subscription.
The Mist dashboard provides a basic level of visibility even without a full Marvis license, allowing administrators to see that certain high-level categories (such as AP, Switch, or Gateway) have identified issues. This is designed to demonstrate the value of the AI engine. However, when you see a lock symbol next to sub-categories or specific Marvis Actions, it indicates that the organizational account does not have an active Marvis VNA subscription for those specific devices.
Without the subscription, the system will not permit the user to "drill down" into the root cause analysis, view historical event timelines, or access the specific remediation steps suggested by the AI. For example, while a standard Wi-Fi Assurance license might show that an AP is offline, the Marvis VNA subscription is what enables the detailed investigation-such as determining if the offline status is due to a cable failure, a bad PoE port on a switch, or a firmware mismatch.
To unlock these features and remove the lock symbols, a SUB-VNA (Virtual Network Assistant) subscription must be applied to the access points in the site. Once licensed, Marvis can provide full "conversational AI" capabilities, allowing you to ask natural language questions about the APs and giving you the interactive tools needed to resolve the issues identified in the Actions dashboard. Therefore, the presence of the lock icon is a direct indicator of a missing or expired subscription tier required for advanced AI-driven troubleshooting.

In a Juniper Mist AI deployment, achieving high-fidelity location accuracy is entirely dependent on the digital twin (the floor plan in the Mist UI) accurately reflecting the physical environment. While scaling the map is the foundation, AP orientation is the most critical variable for the vBLE (virtual Bluetooth Low Energy) engine to correctly calculate a client's position.
Mist Access Points, such as the AP43 or AP45, feature a patented 16-element directional vBLE antenna array. This array does not broadcast in a simple omnidirectional circle; instead, it uses eight distinct directional beams to blanket an area. The Mist Cloud uses "probability surfaces" to determine a user's location by examining the RF fingerprints from these specific beams. If an AP is physically mounted with its LED pointing toward a specific hallway, but is oriented in the Mist UI facing a different direction, the location engine will receive signal data that does not match its internal map of the beams. This results in "flipped" locations or "teleporting," where a user appears on the opposite side of the AP from where they actually are.
To ensure accuracy, administrators must use the small green dot on the outer ring of the AP icon in the Mist Live View. This green dot represents the physical LED on the AP hardware. The orientation in the UI must be rotated so that the green dot points in the exact same direction as the physical LED relative to the floor plan's perspective. It is a common misconception that APs must face true north (A) or geographical north (D); the map and APs can be oriented in any way that is convenient for the facility layout, provided the relative alignment between the virtual icon and the physical device is identical.

To review and troubleshoot the roaming activity of a specific client within a Juniper Mist environment, administrators rely on the conversational and analytical power of the Marvis Virtual Network Assistant (D) and its underlying Marvis Query Language (C). While other Marvis components like Actions focus on organization-wide health and Minis focus on synthetic testing, the VNA and MQL are specifically designed for deep-dive investigations into individual client experiences.
The Marvis Virtual Network Assistant (VNA) serves as the primary interface for this task. It allows an administrator to interact with the Mist AI using natural language. By typing a request such as "Troubleshoot client [Client Name]" or "How was the roaming for [Client Name]?", Marvis VNA aggregates all relevant telemetry data-including signal strength (RSSI), AP transitions, and authentication timings-into a simplified, human-readable summary. It provides a visual timeline of where the client moved and whether each roam was successful, slow, or failed.
Behind the scenes, these natural language requests are often converted into or powered by the Marvis Query Language (MQL). MQL is a structured way to query the Mist Graph API. For specific roaming reviews, the ROAMINGOF clause is used to filter the massive Mist data set for events related only to that client's mobility. For example, a query like roaming of "User-1" during last 24 hours will return a detailed data set showing the "From AP" and "To AP" details, the delta in signal strength during the handoff, and any associated latency.
By using these two features together, administrators can distinguish between a "sticky client" (one that refuses to roam despite a better signal being available) and a "coverage hole" (where the signal drops too low before a new AP is detected). This AI-driven approach eliminates the need for manual packet captures or log parsing, significantly reducing the Mean Time to Repair (MTTR) for mobility-related issues.

In 802.11 wireless networking and the Juniper Mist ecosystem, Passive Scanning is one of the two primary methods a client station (STA) uses to discover available wireless networks.4 In this mode, the client radio does not transmit any data to find an Access Point. Instead, it "waits and listens" on each supported channel for a specific period of time to hear Beacon Frames.5 Beacons are management frames sent periodically by an AP (typically every 100ms) to announce the presence and capabilities of a Basic Service Set (BSS).6 These frames contain critical information such as the SSID (Network Name), supported data rates, security protocols (RSN), and PHY requirements.7 Because the client only needs to keep its receiver active and does not have to power up its transmitter to send probe requests, passive scanning is significantly more power-efficient than active scanning, making it ideal for battery- constrained IoT devices.
However, the trade-off for this power efficiency is time. Since a client must stay on each channel long enough to guarantee it hears a beacon (at least one full beacon interval), scanning the entire 2.4 GHz, 5 GHz, and 6 GHz spectrum can take several seconds. In contrast, Active Scanning (options B, C, and D) involves the client sending Probe Requests (either wildcard/null or directed to a specific SSID) and waiting for a much shorter period for a Probe Response.8 From a Mist AI perspective, the dedicated scanning radio in Mist APs (like the AP43/AP45) performs its own version of passive and active scanning to build a complete map of the RF environment. This data is used by Marvis to detect "Neighbor" and "Honeypot" APs. While most modern mobile clients prefer active scanning to speed up the discovery and roaming process, passive scanning remains a fundamental requirement- especially on DFS channels where regulatory rules prohibit a client from transmitting until it has first
"listened" to confirm no radar is present.

In the Juniper Mist AI ecosystem, troubleshooting the mobility experience of a user requires analyzing the transitions between Access Points (APs). When a client reports a "poor experience when moving," it almost always points to a breakdown in the roaming process, such as sticky clients (devices that refuse to move to a closer AP), slow handoffs, or authentication failures during a roam. The Marvis Query Language (MQL) provides a conversational and structured way to extract this specific data using the ROAMINGOF clause.
The ROAMINGOF clause is designed to filter the massive amount of telemetry collected by Mist and focus specifically on the events surrounding a client's transition from one radio to another. When you execute a query like "Roaming of [Client Name]," Marvis returns a detailed visualization of the client's journey. This includes the signal strength (RSSI) at the time of the roam, the duration of the handoff, and whether the roam was "successful," "slow," or "failed".
By using this clause, an administrator can quickly identify if the "poor experience" is caused by poor overlap (where RSSI drops too low before a new AP is found) or if the client is holding onto a distant AP (sticky client behavior). Other clauses like STATUSOF (A) are used for general health checks, UTILIZATIONOF (C) tracks radio or site traffic levels, and RANK (D) is typically used to identify the best or worst performing entities (like "Rank APs by interference"). Only ROAMINGOF provides the temporal and spatial correlation needed to fix issues specifically related to user movement within the office environment.