Mountain View-based startup Verdigris Technologies looks to do a better job of taking the setup and audit hassle out of building efficiency monitoring. And the company inked a deal for a “significant” installation with San Francisco-based utility PG&E.

Verdigris’ major tool, Energy.AI, applies high-rate sampling to the algorithm that undertakes the task of disaggregation, which is monitoring a building’s AC current draw at an upstream power entry point and identifying every electrical device load in the building by its signature waveform. Hypothetically, this requires a method of decomposing a master-breaker current measurement into a sum of constituent parts, a library of known devices (e.g., Kenmore washing machine Model 110), and an ability to learn or add new or unknown devices. Once the devices and appliances in the building are known, the signature can be monitored for anything that is failing, or aberrant, or simply being left on when it shouldn’t be. The software can communicate with remote handheld devices via an app.

Verdigris was founded in 2010, by Archan Padmanabhan, Craig Norris, and Mark Chung. The firm is a tenant in the NASA Sustainability Base incubator at Moffett Field in Mountain View, California. The company is self-funded with five full-time employees and five part-time or contract employees. Professor Jeremy “Zico” Kolter is a technical mentor. Verdigris has been attracting some attention in startup competitions in the Bay Area, including reaching the semifinalist level in the 2011 Cleantech Open and being judged the winner two weeks ago in the 2nd Keiretsu Forum Pitch Me Green event.

Existing revenue customers include the City of San Jose, the Moscone Center in San Francisco, Blue Earth EMS, and Enovity.

The disaggregation field is not without competition. Sunnyvale California-based Bidgely and U.K. startup Navetas have raised venture funding; Intel has development in-house. And Professor Shwetak Patel of the University of Washington sold his startup, Zensi, to Belkin Corp. for an undisclosed sum. Zensi uses a unique voltage-noise sampling method without relying on current sampling. Like Verdigris, Zensi had never taken outside funding. 

Verdigris similarly is going its own road with regard to sampling method, using high rate sampling with a custom sensor. “We feel that we are unique in what we are doing,” states founder Padmanabhan. “It is refreshing to be in a startup without half a dozen companies right on top of you doing almost the same thing. We believe we have much better discrimination than Bidgely and Navetas. They are using a stock smart meter with low-rate sampling.”

Verdigris is bucking a broad trend of finding creative applications for a standard smart meter.  It remains to be seen whether the cost and installation of a custom power sensor will be a barrier to users.

Farther afield among competitors are conventional building efficiency monitoring players such as SCIenergy, and large corporate entrants such as Siemens and GE.

Setup has been a sticking point for building monitoring software, and has been cited as a rough spot (along with low sales) in a trio of major flameouts last year: Google’s PowerMeter, Microsoft’s Hohm, and Cisco’s Mediator. A disaggregation algorithm will need to overcome messy challenges such as, for example, spotting the twice-yearly load surge from a self-cleaning oven -- and keeping up with the ubiquitous proliferation of novel electronic devices. But if proven robust and reliable, disaggregation could attract users on a new scale.

The built environment consumes 39 percent of North America’s energy. Building efficiency monitoring, if it can translate its potential into reality, may be headed toward being a next layer of “virtual energy supply.”