Fidelix: Creating value with ML and service design

Fidelix, a company at the cutting edge of building automation, offers services that help optimize the energy costs, indoor climate and cost of operations for buildings. This is done by monitoring real-time data from various systems in the building, including air handling units. Scaling up their service business was presenting a challenge, so they contacted us to help them develop their machine learning capabilities. We helped them look beyond their initial understanding of what the problem was and overcome obstacles by using data engineering, machine learning and service design.

Fidelix wanted to have the ability to radically scale up their service business in the future. They came to us to see if we could help them develop their machine learning capabilities to achieve this end, but the direction they wanted to develop their business in was still unclear. We helped figure it out.

The first step on our shared, fully remote journey with Fidelix was to conduct a Current State Analysis in order to pinpoint the problem worth solving. After a series of workshops and interviews, “How to provide valuable, data-driven insights on ten times as many buildings with the same amount of energy experts at our disposal?” was the problem we decided to find a solution for - together.

One of the company’s services is providing building owners and operators with valuable insights about energy expenditure so that they can optimise building automation to increase efficiency. This used to involve energy experts manually processing massive amounts of data produced by various systems installed in the buildings to try and detect anomalies. They were, essentially, looking for a needle in a haystack. We designed and built a tool for Fidelix that would highlight the anomalous needle in the haystack of data and alert their experts. The experts could then evaluate whether the anomaly warranted further scrutiny and could potentially provide added value to the company’s customer.

Analytical results without an efficient way to deliver them to the end-user are close to worthless, so we employed user-centric service design to increase the value of insights gained from the data. To come up with the solutions that would provide value to the experts, we had to almost become an expert ourselves using empathy as a strong and guiding tool to evaluate for example, user interface solutions. We employed customer journey mapping to understand current experience model and to pin point where exactly our solution would have the biggest impact given the scope and time for the project. Different collaboration methods and brutal honesty saved us many hours of meetings to get what it really matters for them done. With a user centered approach we connected user needs to data needs to bridge the gap between those different disciplines while having the business goal as a common denominator. This showed us what multidisciplinarity can bring to the table as it’s best.

Currently, the tool helps energy experts identify and diagnose anomalies with equipment responsible for indoor air quality across multiple buildings owned by multiple clients in Finland.

Automating the part of the process where anomalies are detected allows Fidelix’s energy experts to process much larger amounts of data and use more of their time to provide more valuable insights to building owners and operators. On a larger scale, the project has helped Fidelix start to see data as a strategic asset, helping the company provide it with an appropriate level of focus in their future business.

Building automation has a major impact on the environmental and societal footprint of housing. Improved automation results in improved energy and resource usage in many areas related to living. Air quality as an issue touches individual people, too, and has impacts on areas ranging from quality-of-life to personal health.