It has not too long ago grow to be possible to run private digital assistants on telephones and different private gadgets. On this paper, we describe a design for a pure language understanding system that runs on-device. Compared to a server-based assistant, this method is extra personal, extra dependable, sooner, extra expressive, and extra correct. We describe what led to key decisions about structure and applied sciences. For instance, some approaches within the dialog techniques literature are troublesome to take care of over time in a deployment setting. We hope that sharing learnings from our sensible experiences could assist inform future work within the analysis group.