Performance regressions in mobile apps are difficult to pinpoint, causing significant developer frustration. Datadog collaborated with Google to integrate the new ProfilingManager API from Android 15 into its platforms to address this issue. This API allows programmatic collection of performance data directly from production environments. Datadog now processes millions of production profiles weekly, offering detailed visibility into real-world app performance with low overhead. ProfilingManager enables developers to move from reactive manual reproduction to proactive field analysis of performance bottlenecks. For instance, field traces helped Google identify a background service slowing down app startup. Previously, Datadog's RUM focused on high-level signals, lacking the granular code-level insights needed for root cause analysis. ProfilingManager provides CPU traces, call stack sampling, and memory analysis capabilities. The API offers granular session diagnostics, automated telemetry triggers, and proactive trace snapshots of anomalies. It also enables bottleneck detection at scale across diverse hardware and network conditions. System-enforced resource stability and on-device data controls ensure efficient and relevant data collection. Datadog built a robust infrastructure to process these detailed traces and plans to expand automated bottleneck resolution in the future.