RAN Flow Builder

Building an rApp the old way needs radio expertise and software skills at the same time, plus weeks of back and forth between RAN and dev teams for a single app. RAN Flow closes that gap. A RAN engineer describes the outcome in plain language or composes it on a visual canvas of 22 node types across four categories, then exports a standalone, readable Python rApp. Domain experts ship directly, with no Python, no SDK deep dives and no dev-team backlog. What used to take weeks or months takes days, and each rApp becomes a template the team reuses on its own.

On a live national network, nobody signs off on code they can't read. Every rApp built in RAN Flow is a readable graph, not generated code you trust blind, so engineers open it, inspect it and edit it. Before anything ships, a 12-rule validator and a full dry run against mock RAN data catch misconfigurations, and a human reviews and approves every deployment. Every action ships with built-in guard rails, for example PRB targets held between 5% and 95%, so a workflow can never black out a cell. The result is automation an operator can take through its real change-control process, with an audit trail at every step.

RAN Flow connects to live KPI metrics like Traffic , throughput and PRB usage and uses ML to forecast a KPI 15 minutes to 6 hours ahead with LSTM, ARIMA or your own ONNX model. Workflows branch on the forecast, so an rApp acts before a degradation lands rather than after users feel it. Predictions run inside the same readable graph as the rest of the logic, and every resulting action stays guard-railed and human-approved. On the roadmap: training forecasting models on your own KPI history and versioning them in the platform, so every rApp calls the same governed model.