Blog

Building autonomous networks

Context

Network infrastructure has been reactive since inception. Something breaks, you intervene. Traffic spikes, you provision more capacity. A device misbehaves, you troubleshoot.

“It’s always DNS.”“Have you tried rebooting the switch?”

At Meter, it’s our ambition to ensure these problems never occur. To build networks that are reliable, secure, and now, fully autonomous.

At MeterUp 2025, we shared the next phase of that vision: we’re building autonomous networks. Networks that identify issues early, act before they progress. Networks that maintain performance without manual intervention.

For nearly a decade, we’ve been building the foundation for end-to-end autonomy. It spans the entire stack: the hardware we build, the single operating system running across every device, a sophisticated API, and Command, our generative UI that gives teams a powerful new way to understand and take action on their network.

As networks support more devices, more data, and more business-critical applications, autonomy becomes necessary.

Command unlocks autonomous networks

Autonomous networks are a longstanding industry ambition. The challenge hasn’t been the concept. It’s been the lack of a unified, full-stack network to build them on.

That changed when we launched Command, a small, purpose-built model trained to deeply understand our backend. It allows our customers, partners, and internal team to visualize information, build sophisticated dashboards pulling real-time data, and execute configuration changes on the fly.

Command is already showing how autonomy takes shape with our Support and Operations integrations. In Support, Command analyzes real-time telemetry to guide troubleshooting and resolution. And in Operations, it automates end-to-end network design and configuration, enabling faster, consistent deployments that improve continuously through operator feedback.


        Anomaly Detected:    [] Errors   [] Latency   [] Drops       
   
                                    
                                    
                                    
                
                                                                   
                                                                   
     Routing          Switching            WiFi           Security    
     analysis          analysis          analysis          analysis    
                                                                   
                                                                   
                                                                   
                                                                   
 Review  Verify VLAN Examine RF  Audit ACLs,
     routing         & trunk       parameters        firewall   
    protocols    configurations                    Logs, & IDS  
                    on switch                                   
    OSPF, BGP,        ports                                     
       etc                                                      
                                                                
      Route           Port       Channels, power                
 flaps? errors? interference Anomalies?
                                                                    
                                                                    
                                                                    
 [ ]Yes [ ]Yes [ ]Yes [ ] Yes
                                                            
                                                            
 Recompute Reconfigure Tune Patch
                                                                    
                                                            
                                                            
         
                                    
                                    
                                    
 Validate endtoend QoS metrics, congestion, jitter, etc
                                                                     
 Correlate with realtime telemetry and historical analytics

This tight feedback loop is critical. Each query, each debug session, each network configuration provides raw data that strengthens the underlying models.

Networking deserves its own models

Autonomous networks aren’t achieved by relying on an API call to a generic model. That isn’t how you build systems that can be trusted in production.

Truly autonomous networks require three things: full-stack control, large datasets, and scalable compute.

  • Control: Because we own the full stack, every device, configuration, and telemetry stream is consistent. We can make decisions and take action safely across the entire network. We control the flow of data, rather than having it handed down from predefined systems.
  • Data: Since we design, deploy, and support networks, we have large, clean, and diverse datasets: network designs, configuration changes, telemetry, failures, and fixes, all anonymized.
  • Compute: Through our partnership with Microsoft, we have access to the compute required to train large-scale models on real-world networking data.

With this foundation, we’re training models that deeply understand real-world networks, predict failures before they happen, and fix problems before engineers even know there’s an issue. This is how we built Command, and it's the same foundation that will take Meter from reactive to proactive networks.

Why Autonomy

Network demand is growing exponentially, but workforce growth is flat. Without skilled engineers, we risk security vulnerabilities that stem from misconfigurations and overlooked network design flaws.

We can’t manufacture thousands of new engineers overnight. But we can build systems that help engineers do more. Not to replace them, but to support them. Tools that accelerate design, deployment, and maintenance; improve reliability and performance; and give time back to the people who understand these systems best.

Watch the MeterUp 2025 keynote to hear more on our vision for building autonomous networks.