Founded in 2011, Bidgely uses proprietary machine learning algorithms to “disaggregate” or identify down to the appliance level what’s consuming energy in homes. It then generates personalized recommendations and behavioral nudges to help customers reduce their bills. This provides a duel benefit to utilities who can avoid expensive peak electricity demand periods. Having raised over $51 million through its Series C in January 2018, Bidgely has more than 10 million homes under contract through partnerships with 25 utilities across 12 countries.
Now if you’ve been monitoring the utility industry over the last decade, you might be asking: aren’t the much-hyped smart meters that have penetrated 55% of U.S. homes  thanks to billions in tax & rate payer money  supposed to provide these insights ? The answer is not quite.
Smart meters can only provide home-level energy usage data throughout the day. Bidgely is able to itemize consumers’ energy data into individual appliances by recognizing appliance “signatures” in smart-meter data readings. By leveraging its database of over 50 billion smart meter readings, Bidgely can also apply machine learning to provide itemization to non-smart meter homes. For the latter, a “matched peer” or a “matched region” machine learning model is used. This capability enables Bidgley to offer their solution at a fraction of the cost of hardware-based solutions and to reach 100% of homes, regardless of the types of meters being used.
Currently, Bidgely is focused on launching the third iteration of its disaggregation platform, including support for energy audits and surveys and expanding beyond a mobile app to offer omni-channel engagement. By using its AI capability, Bidgely is able to pre-populate customer surveys which greatly reduces the effort required for completion and improves the accuracy of data collected. Looking forward, CEO Abhay Gupta plans to utilize Bidgely’s latest round of funding ($27 million) to “bring Bidgely’s AI capabilities into even greater focus, increase the velocity of the company’s next growth phase with large utilities, and fuel its ability to scale exponentially in global markets.” 
In 2017, the number of U.S. customers with access to a distributed energy resource (e.g., solar, smart thermostat, battery storage) marketplace ballooned to 32 million and 56 different utilities in the U.S. are operating or planning a marketplace . This increasingly crowded space, though high-growth, puts pressure on Bidgely to offer differentiated services. To best position itself, Bidgely needs to dedicate resources to obtaining more energy data and refining their learning algorithm to identify all possible electronic “footprints.” To achieve the highest level of performance, the algorithms must be “trained on representative datasets that include examples of all possible [conditions].” . Bidgely’s dependency on utility companies for data presents a risk in its business model. Tendril, on the other hand, claims to be able to build customer profiles based not only on utility data, but also on customer and third-party sources.
Bidgely should also move beyond solely providing insights to facilitating automated actions that reduce energy consumption. The average customer thinks about her utility bill just 10 minutes a year ; increasing that number is difficult at best. Bidgely must interconnect with appliances and smart home devices to automatically make adjustments (e.g., turn off lights, modify heating cooling, delay washer / dryer loads) while providing customers an opportunity to opt-out if needed. Dependency on integrating with hardware poses a long-term risk which can be mitigated through merger, acquisition, or partnership. While companies like Tesla are installing battery storage devices (i.e. the Powerwall) paired with solar systems , Amazon and Google are ramping efforts to garner share of home connected devices (including voice assist and smart home products like Nest) . Given the forays of these companies into artificial intelligence [10,11], it will be easier for them to develop capabilities similar to Bidgely’s than vice versa. As such, Bidgely should look into partnerships that enable them to develop their product in coordination with the hardware suppliers to offer customers the most optimal solutions.
Bidgely’s rapid growth and notable investments due to its advanced machine learning capability make it an exciting company to watch. Whether they’ll be able to maintain access to quality energy usage data provided by utilities, automate actions based on their insights, and provide a differentiated product in a world of proliferating interconnected devices remains to be, shall we say, disaggregated.
(Word Count: 750)
 Corpsitedev.bidgely.com. (2018). Bidgely Corporate Website. [online] Available at: http://corpsitedev.bidgely.com/solutions/ [Accessed 13 Nov. 2018].
 Digitalistmag.com. (2018). Connected Assets: How Machine Learning Will Transform the Utilities Industry. [online] Available at: https://www.digitalistmag.com/iot/2018/02/26/connected-assets-how-machine-learning-will-transform-the-utilities-industry-05921360 [Accessed 13 Nov. 2018].
 Cooper, A. (2018). Electric Company Smart Meter Deployments: Foundation for a Smart Grid. [online] Available at: http://www.edisonfoundation.net/iei/publications/Documents/IEI_Smart%20Meter%20Report%202017_FINAL.pdf [Accessed 13 Nov. 2018].
 Businesswire.com. (2018). Bidgely Closes $27 Million Series C to Grow Disaggregation Solutions for Utilities. [online] Available at: https://www.businesswire.com/news/home/20180116005769/en/Bidgely-Closes-27-Million-Series-Grow-Disaggregation [Accessed 13 Nov. 2018].
 Greentechmedia.com. (2018). Utility-Branded Marketplaces Offer DER Services to 32 Million Metered Customers. [online] Available at: https://www.greentechmedia.com/articles/read/utility-marketplaces-der-services-customers-gtm-research#gs.i8WtSDw [Accessed 13 Nov. 2018].
 Grzywaczewski, A. (2018). Training AI for Self-Driving Vehicles: the Challenge of Scale | NVIDIA Developer Blog. [online] NVIDIA Developer Blog. Available at: https://devblogs.nvidia.com/training-self-driving-vehicles-challenge-scale/ [Accessed 13 Nov. 2018].
 Mezger, S. (2018). CUSTOMER CENTRICITY: MUST-HAVE OR A WASTE OF ENERGY? (ACCENTURE). [online] Available at: https://www.accenture.com/t20180612T094240Z__w__/pl-en/_acnmedia/PDF-65/Accenture-Customer-Centricity-Must-have-Waste-Energy-POV.pdf [Accessed 13 Nov. 2018].
 Smart Energy Homes Provide Multiple Grid Benefits. (2017). [online] Available at: https://www.tesla.com/sites/default/files/pdfs/en_US/Tesla_SCE-Powerwall%20Case%20Study-2017.pdf [Accessed 13 Nov. 2018].
 Wang, F. and Holden, C. (2018). How Utilities Are Combining Voice and Analytics for Customer Engagement. [online] Greentechmedia.com. Available at: https://www.greentechmedia.com/articles/read/utilities-explore-customer-engagement-via-voice-and-analytics#gs.783DKVU [Accessed 13 Nov. 2018].
 Tesla (2017). Solar + Powerwall Support Distribution Capacity. [online] Available at: https://www.tesla.com/sites/default/files/pdfs/en_US/Tesla_PG%26E-Case%20Study-2017.pdf [Accessed 13 Nov. 2018].
 Utility Dive. (2017). UK’s National Grid eyes Google AI unit to balance power supply and demand. [online] Available at: https://www.utilitydive.com/news/uks-national-grid-eyes-google-ai-unit-to-balance-power-supply-and-demand/437981/ [Accessed 13 Nov. 2018].
 P. MacDougall, A. M. Kosek, H. Bindner and G. Deconinck, “Applying machine learning techniques for forecasting flexibility of virtual power plants,” 2016 IEEE Electrical Power and Energy Conference (EPEC), Ottawa, ON, 2016, pp. 1-6.