Energymate

Figura 1. Organismos participantes

2. What tasks have been carried out?

To achieve the objectives of the project, a work plan was laid out over the years 2017 and 2018, which highlights the following:

• Study of the state of the art internationally for similar solutions, with a focus on analyzing similar work from leading universities.

• Technological surveillance.

• Development of technical and functional system documents.

• Application of different analytical models on real consumption data from a significant number of test customers provided by EDP, where many conclusions were drawn about relevant and interesting information to be shown to the end user.

• Development of a prototype that clarified the final goal of the platform.

• Development and launch of a first cloud prototype capable of efficiently and scalably processing input data files from EDP for a controlled sample of users.

EDP used for the first time in Asturias the technique known as Pretotyping, developed by Stanford University professor and former Google executive Alberto Savoia, which EDP personnel were introduced to firsthand through the EDP immersion program in Silicon Valley developed in 2017 and 2018.

The prototype is an artifact that will allow us to materialize initial hypotheses to learn as much as possible about the customer, in the shortest time and with the least possible expense. The approach from the perspective of the validated learning cycle (build-measure-learn) will allow us to gather information to support decision-making, incorporating the customer from the earliest phases of service conception (user-centered innovation).

Rooted in the Lean Startup philosophy, this approach aligns with the Lean program implemented in EDP over 10 years ago. The convergence of both methodologies is much simpler in environments where work is already focused on delivering value to the customer, as well as continuous efforts for efficiency and "time to market."

The application of Customer Development principles and Agile development (core pillars of Lean Startup) allowed for prioritizing the validation of initial hypotheses (customer interest) before making a larger investment by the company, which might not yield direct returns in perceived value by customers.

Once the prototype was made, the project aimed to achieve a series of functional blocks based on a technological architecture over AWS infrastructure.

3. What were the stages of the project?

The main stages completed in the project were:

• Data collection.

• Functional analysis and technical design.

• Modeling.

• Back-office implementation.

• Front-office implementation.

• Stabilization, testing, integration, and deployment.

• Cross-functional phase of prototyping and validation.

4. What results have been obtained?

The results obtained were especially focused and deepened in the following points:

• Prototype: It has been a valuable source of information for consolidating the project. It has strengthened the knowledge we have of the market and, more specifically, of energy customers. The conclusion drawn from it is that the user wants to be well-informed to make the most appropriate decisions regarding their energy efficiency. To reach this point, there has been an increase in efforts on social media and web content to create a community of users interested in energy efficiency.

• Analytical and Data Hybridization Module: Initially, the analysis and development focused on this point, resulting in the creation of different models and promising preliminary results, linking electricity consumption with weather conditions and holiday calendars.

• Natural Language Generation (NLG) Module: This module, along with the next one, became the true focus of the project, receiving more attention than initially planned. In this module, the generation of natural language from the consumption data files provided by EDP was enhanced.

• Decision Module: Designed for greater flexibility and ease of use.

• Interaction Application (Front-office): To monitor the production of messages and the processing of input data files from EDP.

Figura 2. Esquema de funcionamiento de la aplicación

A Gamification Module is pending for later phases, which will complete the user profiling mechanisms and the customization of messages for the recipients.

Figura 3. Aplicación Energy Mate

Figura 4. Ejemplo de mensaje generado

5. Value proposition offered. What advantages does it present over other technologies?

It provides customers with relevant information about their electricity consumption habits using natural language from their readings, in an innovative service, with the ability to differentiate itself and thus retain customers by fostering loyalty and strengthening relationships with them. It transforms numerical data in kWh into explanatory messages about electricity consumption.

The difference from the competition lies in combining three key innovative aspects in a single solution:

• Consumption data analytics based on smart meter measurements.

• Application of natural language processing to transform numbers into easily understandable messages.

• Use of gamification techniques to “engage” consumers with the solution and provide information about their characteristics to improve the experience and messages generated about consumption.

The main values for the customer include a better understanding of their consumption habits, greater empowerment by understanding how, when, and why they consume, and mechanisms to help end consumers improve their consumption to save energy and achieve greater energy efficiency.

This can be summarized as: “Energy Mate, your ally to reduce your electricity consumption.”

6. What key points and innovative elements can be observed from this project?

The most relevant elements of the project are:

• Scalability: Processing large volumes of data (Big Data architecture).

• Integrability: Quick transition to production. Ease of integration with corporate back/front offices in different formats and technical requirements.

• Flexibility: Ease of adding new messages and business rules.

• Personalization: Natural language generation module and templates.

• Modular: Modular and atomic design that allows the incorporation of new requirements.

7. What are the next steps for the project?

The following are ideas for the future, new challenges to face in the next phase of the project:

• Increase the number of users.

• Enrich and increase the number of messages.

• New Dashboard with new message metrics.

• Increase capabilities and functionalities for internal promotion (back offices, screens, presentations, etc.).

PROJECT FUNDED BY