there's tooooooo much latent compute in the world today. In your phone, laptop, desktop, watch, etcetera. I think extracting this edge compute is very much possible through browsers today. Reduce your server costs, handle your client's data more responsibly -> use their beefy devices to run your ML models