Meter’s approach is different. All hardware runs identical firmware. Telemetry data has the same structure whether it comes from an access point, switch or firewall. This consistency lets Meter train its Command AI model on a clean dataset.
Command now operates in two areas: support and operations. For support tickets, Command analyzes real-time telemetry, runs model jobs and recommends actions. The system currently handles 85% of tickets with model-generated insights. Meter reserves 15% for control group testing.
For operations, Command automates network design. After sales calls where customers describe their setup, Command reads transcripts and generates complete network configurations. This includes topology, architecture, required hardware, VLAN assignments, IP addressing and security policies.
Desired state configuration takes flight
Traditional network configuration is generally imperative. Engineers log into switches and routers to run commands that change device state. Each configuration change requires CLI access or API calls to individual devices. Configuration drift happens when devices get out of sync.
“All of the source of truth and intelligence actually lies in the back end, and this is what we call desired state networking,” Varanasi said.
The architecture builds on software-defined networking (SDN) concepts that have been in the market for over a decade and abstracts them out further. The source of truth lives in the cloud, not on individual devices. Engineers define what the network should do. Meter’s system figures out how to configure hardware to achieve that state.
