neuralSPOT SDK for Edge AI Development

neuralSPOT® supports Ambiq Apollo generations 3 through 5, including Apollo3, Apollo4, and Apollo510 series SoCs. Support is continuously expanded to future Apollo generations as new devices are introduced. 

neuralSPOT® provides first-class support for TensorFlow Lite for Microcontrollers (TFLM), heliaRT™, and heliaAOT™, allowing developers to choose the AI inference runtime that best matches their model size, performance, latency, memory footprint, and deployment requirements. 

neuralSPOT® powers several application-focused AI development kits, including: 

• heartKIT™ for ECG-based heart rate, heart rate variability (HRV), and atrial fibrillation (AFib) detection.  

• physioKIT™ for optical-sensor-based SpO₂, heart rate, AFib detection, and cuffless blood pressure estimation.  

• sleepKIT™ for real-time, on-device sleep monitoring and sleep-stage analysis.  

neuralSPOT® includes ready-to-use AI pipelines for keyword spotting, speech-to-intent, speech enhancement, and speaker identification. These capabilities enable battery-powered devices to perform voice processing locally, reducing latency, improving privacy, and minimizing cloud dependence. 

Yes. neuralSPOT® is a fully open-source, OS-agnostic SDK that developers can inspect, modify, and extend. It is designed for real-time edge AI on ultra-low-power embedded systems and supports customization for production applications. 

No. neuralSPOT® runs on supported Ambiq Apollo SoCs and development boards. Developers can prototype, profile, optimize, and deploy AI models directly on Ambiq hardware using plug-and-play PC integration and the included development tools. 

Yes. neuralSPOT® is designed for on-device AI inference, allowing machine learning models to execute directly on Ambiq Apollo SoCs without requiring cloud connectivity. Running AI at the edge reduces latency, improves privacy, lowers power consumption, and enables reliable operation even when internet access is unavailable.