NetBeez Presents at Networking Field Day 41

From Troubleshooting to Continuous Network-Experience Assurance with NetBeez

Watch on YouTube
Watch on Vimeo

NetBeez was built around a simple idea: understand network performance from the user’s perspective. In this session, Stefano will introduce NetBeez, explain how the company and platform have evolved since its previous Networking Field Day appearance, and share the vision behind NetBeez NextGen. The focus is the shift from primarily reactive troubleshooting toward continuous network-experience assurance, helping teams detect degradation, isolate the likely fault domain, take action, and verify improvement both during and between incidents. Stefano Gridelli, co-founder and CEO of NetBeez, presented the session at Networking Field Day 41. It was the company’s first appearance since Networking Field Day 28, about four years earlier.

Gridelli explained that infrastructure monitoring tools can show green dashboards while users struggle with Wi-Fi, VPN connections, or slow applications. NetBeez addresses this by deploying distributed agents that test continuously from the user’s vantage point. The agent lineup now includes branch appliances and software agents, Wi-Fi sensors, cloud and data center agents, and lightweight Windows and Mac clients for remote workers. NextGen organizes the data into a four-phase loop of detect, correlate, act, and prove, and uses moving averages to catch gradual degradation. A redesigned main dashboard highlights where attention is needed. It includes cards for open incidents, a geolocation map, SSID health, ISP connectivity, and application performance, along with remote users grouped into problem buckets such as gateway or VPN issues.

In response to delegate questions, Gridelli clarified that NetBeez reports problems rather than fixing them, since it does not integrate directly with network hardware. The same data that detects a problem also shows whether a fix worked, and API and MCP integrations help teams close that loop. Custom dashboard views will let Wi-Fi, desktop support, and infrastructure teams focus on their own domains. LLM-driven summaries are meant to cut the time needed to digest the data. Gridelli also confirmed that agents still run on any Linux or Windows hardware, including Raspberry Pi. By emphasizing proof of improvement between incidents, NetBeez aims to help teams catch slow degradation and validate changes before issues reach the help desk.

Personnel: Stefano Gridelli

How NetBeez Agents Turn Continuous Measurements into Fault-Domain Evidence

Watch on YouTube
Watch on Vimeo

Panos takes a hands-on look at how NetBeez continuously measures network experience across branch offices, Wi-Fi networks, and remote workers. Using the NextGen interface and Buzz dashboard, he will show how teams can identify degraded experience, understand emerging and recurring issues, and isolate the likely fault domain using evidence collected from the user’s perspective. The live scenarios will include a Wi-Fi access point failure and a shared VPN-path degradation affecting remote users. NetBeez co-founder Panos Vouzis led the demonstration at Networking Field Day 41.

Vouzis described the full agent lineup. It includes Raspberry Pi-based appliances, VM and container agents hosted on SD-WAN equipment or in the cloud, hardware Wi-Fi sensors, and remote worker agents for endpoints. All of them run the same synthetic tests, covering latency, jitter, packet loss, HTTP response time, VoIP quality scoring, and speed tests. Vouzis said the Wi-Fi sensor is hardware by design: a stationary, identical device allows objective comparisons across days and sites and also reports RF conditions. In the first scenario, a wireless-only sensor in the Barcelona office showed latency and packet loss spiking around 10 a.m. At the same moment, its signal strength fell as it roamed from the nearby access point to a distant one. A Wi-Fi scan from the sensor found only one of the office’s two access points broadcasting, which confirmed the failure. In the second scenario, the VPN card flagged three remote users. Their round-trip times held steady while packet loss reached about 10 percent, which pointed to the VPN concentrator.

Delegates raised concerns about agent sprawl. Vouzis said campus customers typically rely on hardware sensors, while endpoint agents suit smaller groups such as call center staff, traveling attorneys, and executives. He added that the endpoint agent uses a fraction of a percent of CPU and roughly 30 to 40 MB of memory. NetBeez offers both SaaS and on-premises deployment and supports customer-managed encryption keys. It can retain data for up to two years and can trigger alerts from deviations against a custom baseline. For zero-trust environments, Vouzis suggested placing an agent at the VPN infrastructure alongside the laptop agent to get an end-to-end view. Both scenarios ended the same way: once the fix was applied, the same continuous measurements would confirm a return to baseline.

Personnel: Panos Vouzis

From Troubleshooting to Configuration Through Natural Language with NetBeez

Watch on YouTube
Watch on Vimeo

Panickos will demonstrate how AI is being integrated into NetBeez NextGen to help operators troubleshoot problems and manage monitoring using natural language. The session will cover the NetBeez AI architecture, including support for NetBeez-hosted AI and customer-selected AI providers. Live demonstrations will show an L1 operator using AI to investigate and summarize a user issue, create a monitoring target, and tune alert profiles, as well as how MCP enables NetBeez data and capabilities to participate in broader AI workflows. Panickos Neophytou of NetBeez closed the company’s presentation at Networking Field Day 41.

Neophytou described three ways AI fits into NextGen: answering open questions from NetBeez evidence, testing an operator’s hypothesis on demand, and turning intent into actions such as creating targets. The NextGen server exposes purpose-built tools for querying telemetry, running ad hoc tests, and changing monitoring configurations. Operators reach these tools through an in-dashboard assistant or through an MCP server. The dashboard assistant inherits the context of the current view, including the agent, time window, and tests on screen. Neophytou said the MCP server is integrated with the platform rather than wrapping the API, and that everything it offers is also available through a JSON API. Customers can bring their own API keys for any AI provider, use a model hosted by NetBeez, or point the assistant at a locally run model. In the demonstrations, the assistant summarized the day’s priorities from the main dashboard. Without being told the cause, it traced degraded call quality in the Barcelona office to Wi-Fi instability on a single sensor. It then requested approval before running ad hoc ping tests to techfieldday.com from five agents and created a monitoring target from the results.

Delegates asked about the direction of data flow with AIOps platforms. Neophytou said that through MCP, NetBeez becomes another data source and configuration endpoint that customer-built agents can query and act on alongside documentation and other tools. Those agents can also generate standalone dashboards from the NetBeez API. He confirmed that operators can ask the assistant to recheck an issue after making a fix, which closes the verification loop that the earlier sessions described. Looking ahead, NetBeez plans to add more tools and a background agent that continuously analyzes data and delivers scheduled reports. Neophytou stressed that operators stay in control and that NetBeez is not trying to build another black box.

Personnel: Panickos Neophytou