http://www.postscapes.com/wifi-home-energy-monitor-neurio/
On Hacker News, this one (Open Source and Raspberry PI based) would probably be more relevant:
In our case, we are focusing on reliable device detection of major home appliances (we detect home appliances which consume more than 100W). We detect devices by our algorithms and users can teach the system via training mode for more precise recognition. Beside this, we support 3-phase electrical networks (for both grid and grid+solar) and we support 220V/50Hz networks (so Ecoisme works in both Europe and US).
And - at least to me - it would be interesting if (without of course revealing any trade secret) you could describe how you can identify devices, i.e. which parameters allow you to distinguish (or classify) devices connected to the same breaker.
Something like this (now old) paper? http://abstract.cs.washington.edu/~shwetak/papers/ubicomp200...