Leveling Up Building Operations Through Generative AI

With the breakthrough of artificial intelligence, the building sector is on the verge of a profound transformation. By combining data analysis, actionable recommendations, and predictive control, this technology already offers entirely new possibilities for improving resource efficiency and reducing the carbon footprint of buildings.

But what if there were a digital building assistant that not only analyzes information but also develops new solutions for the constantly changing challenges of operations and responds in real time to inquiries from owners and operators? With the implementation of generative AI, this is no longer a distant utopia, but already a reality today.

Why do we need Artificial Intelligence in Buildings?

Since the breakthrough of ChatGPT, it has become clear: artificial intelligence (AI) will fundamentally change the way we work — and the real estate industry is no exception. Buildings are becoming increasingly digital and complex, and their efficient use and sustainable management require extensive expertise and a high level of responsiveness. AI can assist in this context by

• analyzing large volumes of operational data in real time,
• conducting predictive analyses,
• providing tailored recommendations.

The applications extend across the entire lifecycle of a property: from optimizing materials and building designs to demand-oriented and predictive control of heating, ventilation, and air conditioning (HVAC) systems, as well as the development of new resource concepts in line with circular economy principles and the energy transition. One example is demand side management, where AI strategically controls the electricity consumption of buildings to reduce costs based on market signals and to maximize the potential of renewable energy sources.

Generative AI as a Productivity Booster

Generative AI (or “Gen AI”) takes the user experience to a whole new level: While traditional AI analyzes, categorizes, and summarizes existing data to solve specific predefined tasks, generative AI goes a step further by generating new content and solutions from existing information and spontaneous inputs — content that did not exist before.

This includes texts, photorealistic images, videos, and audio content. Such capabilities are made possible, in part, by the use of Large Language Models (LLMs), which can contextualize words and produce coherent text. Multimodal Large Language Models (MLLMs) expand these capabilities by processing and generating various data types, such as text, images, and audio files simultaneously.

This ability makes generative AI a true "productivity booster". According to a study by the consulting firm McKinsey, generative AI technologies could enable a global annual productivity increase of approximately 2.4 to 4.1 trillion euros – an increase that significantly expands the boundaries of what is possible, even in the context of building management.

Applications of Generative AI in Buildings

So far, AI in buildings has focused on a limited number of predefined tasks based on coded rules and algorithms: structuring and analyzing building data, adjusting setpoints of HVAC systems, detecting malfunctions, recommending actions, or implementing predictive controls based on weather forecasts. While these traditional applications are extremely useful, they are limited in their adaptability. Generative AI, on the other hand, enables a more tailored response to the specific requirements of properties and accelerates processes. Here is how it works:

Training the Model with Extensive Data Sources

The model is initially trained by the respective software provider using vast amounts of data from public sources, such as technical documentation, building plans, sensor information, and other databases. For instance, thousands of operational manuals from all major building automation manufacturers, as well as scientific publications and case studies, can be included in the training set. This wide variety of data sources gives generative AI a comprehensive expertise that far exceeds what conventional AI models offer.

Interactive Communication via Chat or Voice Input

Based on this information, the digital building assistant can respond to so-called “prompts” – input requests in everyday language. Through an intuitive user interface, such as a chat window or voice input, facility managers and technical operations teams can ask questions or provide specific instructions, such as: “The temperature setpoint for the conference room is not being reached as desired. What should I do?”

Tailored Recommendations for Operational Optimization

Immediately upon receiving the question, the system accesses the stored knowledge about the building, activates the relevant information from the training data, and offers tailored suggestions for optimizing operations. It first searches real-time sensor data, maintenance logs, and user manuals to identify the root cause of the problem. In our example, the generative AI might determine that a faulty outdoor temperature sensor is likely to be the cause. The result is a precise recommendation in written form or natural language, such as: “Ensure that the outdoor temperature sensor isfunctioning properly.”

The answers are tailored exactly to the conditions in the building. Additionally, when it receives feedback on its answers, the digital building assistant can incorporate this new information contextually into its future responses to a certain extent.

But it is not just technical operating teams that can benefit from Gen AI. The technology also provides owners and asset managers with a simple and quick way to stay informed about the performance of their portfolio:

Key Technology for Efficient Buildings

These examples demonstrate that generative AI is revolutionizing the way buildings are managed and operated. Asset and facility managers can enhance their productivity through simple inputs, retrieving information in natural language, testing hypotheses, and receiving immediate support for their everyday decisions. The benefits are clear:

• Tailored responses to optimize each property in the portfolio
• Shortened response times to error notifications and operational inefficiencies
• Reduced administrative effort
• Improved user experience through intuitive communication via chatbots or voicebots

The diverse applications of generative AI open up entirely new pathways for reducing energy consumption, operating costs, and CO₂ emissions within the portfolio, providing optimal solutions for future challenges – more efficiently, sustainably, and successfully than ever before.

‍

‍

‍