Skip to main content
European Climate, Infrastructure and Environment Executive Agency


Personalised virtual assistant boosting energy efficiency at home.


Eco-Bot’s main objective is to engage consumers to change their behaviour towards more energy efficiency practices, which is a key factor in the endeavour of tackling climate change. Changing energy behaviour is crucial as it is an important factor influencing the overall energy consumption and energy efficiency.

The Eco-Bot project unveiled an innovative system that motivates consumers to manage energy more efficiently by providing personalised information on energy consumption using a chatbot.

An important element of the Eco-Bot approach is a multi-factorial behavioural model, which uses data mining and segments users according to their motives to save energy and their behaviour related to energy saving and environmental protection, classifying them into ecological idealists, aspiring ecologists, dedicated savers, opportunists, and indifferent. The behavioural model enables the provision of tailored recommendations for each of the segmentation sectors, so as to properly induce consumers to save energy and to promote pro-ecological behaviour. For even more targeted recommendations, the behavioural model is also combined with additional factors such as the user’s income and type of residency.

In addition to the personalisation offered through the developed multi-factorial behavioural model, energy disaggregation algorithms were also deployed in order to provide appliance-level consumption information and itemized billing, thus further supporting Eco-Bot users in becoming more aware of their consumption habits and realising ways to improve their energy efficiency. The non-intrusive load monitoring (NILM) module that was developed is based on NILM algorithms that operate on low sampling rate smart meter type data. In order to meet the varied reliability, sampling and scalability constraints of the user groups and use cases where no sub-metering data is available for training, the NILM algorithms that form the basis of the NILM module include novel supervised, transfer learning based appliance-specific NILM models for different residential pilots’ resolutions, as well as prediction-based commercial NILM models.

The Eco-Bot platform integrates the behavioural model that enables to assign the user to the appropriate segment and provision of relevant, personalised recommendations, and the NILM module that performs a disaggregation of consumption data at appliance level. The platform also enables the integration with energy management systems and energy service provider platforms, allowing the collection of smart meter data of residential customers as well as of tertiary buildings.

Through an attractive front-end interface, the chatbot enables seamless communication in a more natural and interactive way than that offered by traditional mobile applications. Eco-Bot enables users to make a wide range of inquiries and gain access to their own energy data, providing detailed information on consumption at appliance level, monitoring and comparison of consumption (total or appliance-specific) in different periods, energy and cost savings achieved after the implementation of energy efficiency improvements, etc. Moreover, overconsumption alerts and other notifications help in keeping users aware and alerted of their consumption.

Eco-Bot was demonstrated and validated through three large scale pilots across Europe representing different business models (B2C, B2B2C, B2B), involving consumers and facility managers from Germany, Spain, Italy, and the UK. Despite the fact that the Covid-19 pandemic affected the energy consumption of the users in all pilots, the pilot findings showed that Eco-Bot motivated both residential and commercial users to improve their energy consumption behaviour both by implementing the Eco-Bot recommendations and making use of the suggestions to invest in more energy-efficient appliances.

According to the findings, Eco-Bot positively influences the energy consumption of users compared to the group of non-Eco-Bot users, as Eco-Bot users showed a reduction or a lower increase in energy consumption in the Covid-19 era. Moreover, the energy efficiency targets for the commercial buildings pilot were reached, despite the shift in priorities of energy managers due to Covid-19. It should be noted that, although the project did not focus on renewables, a survey conducted after the completion of the pilots, showed that 80% of the residential users were interested in either switching to renewable energy plan/provider or producing/supporting renewable energy themselves, after their experience and knowledge gained through Eco-Bot.

CORDIS project factsheet

CORDIS project Results in Brief

Eco-Bot website

Eco-Bot YouTube channel




Xantener Strasse 11, 10707, Berlin, Germany