Privacy is a legitimate concern, but there are useful things you can do that you might feel comfortable sharing with our servers. Zero-information computing is really hard, and we didn't feel it was worth the time it'd take to give that kind of strong guarantee.
The hardware is open. If you want to run your own software on it you can. We did have a big debate whether to go the zero knowledge route (we previously worked on building a zero knowledge file service).
Seems quite a marketing lie, zero knowledge, while you really want _all_ the knowledge about all my sensors, and the way I use your product. While I get no knowledge of how your servers work and what you do with it.
We make no claims of being architected Zero Knowledge.
We're familiar with the concept from previous work, but decided it wasn't worth creating an MVP with it.
One can architect a service to be zero knowledge like the terrific job http://www.tarsnap.com/ does. Unfortunately it's pretty hard for us to do when the core value proposition of our product is that we connect your hardware to other APIs (Dropbox, Facebook, Twitter etc) (which don't follow a zero knowledge architecture).