68% of mobile operators say they have an AI strategy for their networks.
But when we look at their daily operations we always see the same pattern: dozens of dashboards, siloed tools, engineers correlating data across systems that don't talk to each other.
The gap between strategy decks and daily operations is enormous.
Next week we'll be at FutureNet World in London (April 21–22), one of the few events focused specifically on network automation and autonomous operations. We'll have a stand there.
Here's what we'll be talking about.
The tool problem
Network operations today run on a model designed twenty years ago. Operators use hundreds of specialized tools for planning, troubleshooting, optimization, monitoring. Each one covers a specific use case with its own data, its own interface, its own logic.
An engineer investigating a performance issue needs to check counters in one system, pull call traces from another, cross-reference alarms in a third, and manually piece together what happened. This process works. It's also slow, expensive, and entirely dependent on the experience of the person doing it.
Adding AI on top of a single siloed data source gives you a slightly faster version of the same fragmented process.
What actually needs to change
The shift toward autonomous operations requires three things to be true at the same time:
- Unified network data. PM counters, call traces, configuration parameters, alarms, topology, geolocation — decoded, correlated, and available across use cases. One fabric, decoded once, usable everywhere.
- Telecom-specific algorithms operating on that data. Anomaly detection, interference classification, congestion prediction, root cause analysis. These require deep domain knowledge of how mobile networks behave. Generic ML models don't get you there.
- AI agents that execute real operational tasks. Agents that analyze issues, identify causes, and recommend or execute corrective actions. The engineer defines the intent ("understand why dropped calls increased in this area") and the system does the rest.
When these three layers work together, you move from operations driven by human interpretation to operations driven by data and automation.
Where we are in this
At Kenmei, this is what we build. Our platform integrates multi-vendor, multi-technology network data into a unified Telco Fabric.
On top of that, we run specialized algorithms developed for RAN analysis: interference detection, signaling troubleshooting, geolocation, traffic segmentation.
And on top of that, we deploy AI agents that coordinate these capabilities to complete operational processes with minimal human intervention.
We've been doing the data and algorithms part for years. The agents layer is where things move now, because it closes the gap between analysis and action.
If you're at FutureNet World, come say hello.
