HeatTransformers, founded in 2018, had the economic wind in its sails for a long time, but heat pumps have since found themselves in different weather. Due to a range of factors, strong growth rates have declined. This spring, moreover, the obligation to have central heating boilers replaced by (at least a hybrid) heat pump from 2026 was scrapped. 'That's unfortunate,' said De Ronde, who set up the data and BI capabilities at HeatTransformers and supervised several strategic data-related projects. 'But fortunately, there are plenty of positive drivers for purchasing a heat pump. It is a sustainable option and, just as importantly for many people, it helps save energy costs from a certain gas usage. For some, geopolitical motivations come on top of that.'
The fear of the unknown is fading, according to De Ronde: 'You hear less and less about heat pumps making noise, looking unattractive, or malfunctioning. Recent research by the RVO proves that owners are quite satisfied with them: 95 percent recommend a heat pump.'
“ We are not monitoring in a laboratory setting, but in a home setting. And the more homes, the better we can optimize.”
Complex models for improved prediction
The market prospects for the slightly longer term are therefore fine, as indicated by De Ronde, but even in these less favorable times HeatTransformers is doing well: the multi-site company is even among the 250 fastest growing companies in the Netherlands. Their data-driven approach is paying off: 'We make a lot of use of smart data insights,' says De Ronde. 'Specifically: complex models to better predict, from customer conversion rates to lead times, drawing on various data sources. We remotely gather data from approximately five thousand of the heat pumps we've installed. We have defined parameters to chart the success of the pumps. So we are not monitoring in a laboratory setting, but in a home setting. And the more homes, the better we can optimize.'
New AI applications
In addition to machine learning ("old-fashioned AI, you might say"), De Ronde sees new applications of artificial intelligence making their appearance: "Agents that take over tasks, based on LLMs (large language models, ed.) like ChatGPT. Think of tools for customer service, automatic note-taking et cetera. All the analysts and developers here have ChatGPT open to improve codes. ‘The developments are advancing so rapidly, the use cases are popping up like mushrooms, but it's not easy to implement them yourself.’
HeatTransformers gained experience with ‘an internal chatbot on its own knowledge system,’ De Ronde continued. That should be able to give customers concrete suggestions based on housing specifications. "We tested two applications, but it turned out to be complex to get it error-free. You can say that the technology did the 80 percent well that we can already do well ourselves. But the added value is right out there. What we noticed was that the application always gave an answer, while in some cases no answer was possible, the well-known hallucination. Especially when it came to specific properties of heat pumps. It keeps on testing and learning. In our core processes, we certainly still need people for the time being.'
More than a push of a button
De Ronde is brainstorming with colleagues on how HeatTransformers can best implement new capabilities into its business model. 'We're not just talking about one push of a button, but about entire projects that require a lot of attention. To start with, your data infrastructure must be in good order, because it is difficult to extract smart AI from bad data. In addition, you have to know very well what exactly you want to generate. Standardizing and getting teams on board takes time, you shouldn't think too lightly about that. I also hear this from other data-driven companies. I am convinced that organizations that are currently using advanced AI applications have invested a lot of time and tested in silence for a long time before they started using them. It takes a lot of small implementation steps to achieve big dreams in AI.'
“ There are significant challenges to solve grid congestion and to use data to optimize energy use and storage, at both household and neighborhood levels.”
Data Expo and inspiration
At Data Expo in September, De Ronde hopes to gain new insights, realizing that data will become increasingly important in the energy market: 'There are significant challenges to solve grid congestion and to use data to optimize energy use and storage, at both household and neighborhood levels.' But there are also lessons to be learned beyond the industry, according to the Strategic Project Officer. Last year, De Ronde was inspired by a presentation by the Municipality of The Hague, on linking clients to WMO specialists. 'Although it is a very different type of organization, we found ourselves facing exactly the same constraints. Like them, we strive to make matches as effectively as possible, between customers and installers in our case. I find it especially interesting at Data Expo to hear analysts, data scientists and developers talk about concrete cases. For instance, what challenges have they overcome, what solutions have they chosen, and what possibilities and limitations do these present?'
_____
At the upcoming Data Expo, there will be a strong focus on AI. What opportunities and challenges does artificial intelligence offer for organizations? At the Jaarbeurs on Wednesday September 11th and Thursday September 12th, you can gain essential knowledge, explore technological possibilities, and make valuable connections to take new steps forward. And not just in the field of AI: visiting Data Expo helps you find answers to all your data-related questions. Visitor registration is free and can be done here.
_____
Also read: Getting started with AI? 6 points of interest