Reasonance has already been awarded the innovation prize for artificial intelligence (AI) by the Ministry for Economic Affairs and Climate Action Baden-Württemberg for its platform Atlas. The start-up’s approach combines traditional modeling with machine learning. Atlas offers tools for training, versioning and deploying AI models. The solutions are already helping the energy industry improve PV power generation and consumption forecasts.
We officially founded Reasonance GmbH in May 2020 – in the middle of the COVID-19 lockdown. But its history goes back even further. I was working on my thesis on AI at the Fraunhofer Institute and freelancing for companies in the Netherlands, Israel and in the US. And then one day it hit me: Why not apply machine learning to problems that really matter – such as in the energy, chemical or pharmaceutical sectors? I started by pitching this idea to some of my fellow students. That is how I met the co-founders Konstantin and Manuel. Our first client was a Dutch start-up in the energy sector for whom we developed a real-time optimization system for district heating networks. With the help of our technology, the young company went on to become a global market leader. This inspired us. We got into contact with EnBW’s subsidiary Netze BW in 2020, but to work with them, we had to found a private limited company (GmbH).
In the beginning, we did not have a specific product in mind, but over time we developed a clear vision. Basically, our goal was to make AI more accessible to companies through standardization and repeatable processes. Over the past four years, we have observed that 90 percent of machine learning projects focus on repetitive tasks, such as processing data, creating dashboards, setting up data bases as well as versioning and training models. We realized that we could automate these tasks, allowing data scientists and engineers to focus on their core work. This is exactly what our Atlas platform is for.
To make the platform more tangible, we developed specific AI solutions for the energy industry – including PV power generation and consumption forecasts. We have found our approach of combining traditional modeling with machine learning to be extremely effective. Our energy demand and generation forecasts recently earned the AI-Champions Award from the Ministry for Economic Affairs and Climate Action Baden-Württemberg. Today, we are working with companies in more than ten countries.
We offer software as a service (SaaS) or platform as a service (PaaS) solutions. Although we still sometimes develop customized solutions and offer consulting services, we are reducing that segment. Our core product is the Atlas PaaS solution. It enables companies to effectively use machine learning and data-intensive applications. I know it sounds complicated – but essentially, we help companies use machine learning for repetitive business processes. This greatly increases their efficiency. However, we have found that German organizations are still reluctant to use such platforms. They continue to focus on specific use cases. For this reason, we have recently shifted our focus in the energy sector to small-scale solutions, such as energy demand or solar power forecasts, that deliver value from day one.
In any given project, we always focus on a number of key elements. We need a specific goal and measurable improvements, so we define clear key figures and processes to improve. Our approach is different from others because we implement, measure, evaluate, and adjust in rapid cycles. We create data from the repetitive processes and visualize the progress in what we call the “big book of plots”. We identify two different sets of key figures: one internal set for technical goals and one that is easier to interpret for business stakeholders. Business stakeholders always need to be well-informed and able to understand the key figures.
In fact, we have identified several effective use cases for AI in the energy sector. It can optimize energy procurement and trading, lowering consumer prices and increasing margins. Trading solar power can generate high returns at the energy exchange. Especially in the low-voltage grids, AI can increase transparency and help optimize grid planning in general. What's more, as in other sectors, customer service can be automated through intelligent routing and chatbots. AI can better estimate the service life of critical infrastructure, such as transformer stations, and develop grid flexibility solutions.
Our platform centralizes and optimizes all data-related processes, from collection, processing and analysis to AI model deployment. Atlas enables a seamless integration of data from various different sources, including smart meters, sensors and other energy systems. The platform is capable of handling growing data volumes, which is critical for the data-intensive energy industry. Atlas offers tools for training, versioning and deploying AI models, supporting the entire life cycle of model development.
It also comes with pre-integrated AI apps designed for the energy industry, such as our electricity demand and yield forecast, which helps utilities optimize their resources. The architecture ensures that companies maintain full control of their data while benefiting from AI. Atlas reduces costs by automating many repetitive tasks.
It goes without saying that we want to provide our services to as many utilities and solar farm operators as possible. Our smart meter data management and analysis solution also includes dynamic pricing. We want new customers to quickly understand how to use our platform, so we will improve the onboarding process. Because from January 1, 2025, smart meter data management will be required by law. We also want to work more closely with energy traders to provide solutions for utilities that are unable to trade themselves.
On the technology side, we are currently focusing on combining the two forecasting models with an optimizer for planning energy storage systems. The potential for savings and grid flexibility is enormous. However, operating one system per building or storage system is going to be challenging.
This interview was conducted by Niels H. Petersen.
Are you a start-up, too? Would you like to present your ideas and products for an efficient, clean and reliable power supply to a large, international audience while keeping your costs to a minimum? With its range of packages and numerous free networking opportunities, the Start-up Area at The smarter E Europe in Munich from May 7–9, 2025, could be the perfect solution for you.
