The WebNN service exists to facilitate neural network inference with accessing the hardware accelerated OS machine learning API.

WebNN explainer: https://github.com/webmachinelearning/webnn/blob/main/explainer.md

WebNN spec: https://www.w3.org/TR/webnn/

WebNN service design doc: https://docs.google.com/document/d/1TMs36IE9wL9rNuh8lriGr51S8MU1JezU2SCZy-YqJ3Y/edit#heading=h.3z6obdxyjeka