What If Your Building Could Tell You What It Needed?

Natural language processing. That is the technical term given to the ability for you to ask a computer a question and have it respond to you. It is also one of the technologies behind the fast-paced growth of ChatGPT. By being able to extract important information from a question, no matter how it is asked, and framing the response in plain language, tools like ChatGPT are able to offer the power of AI to the masses. This same capability for building systems would change the way that buildings are managed. 

Right now buildings are already recording a ton of data, but the sheer amount of data being captured makes it a laborious task to understand how the building is actually performing. Imagine if this could be done with an easy question, “Am I reducing my energy and carbon emissions compared to this time last year and if so how much money am I saving?” The AI would understand what data it would need to measure the building’s performance by and summarize it for easy consumption. The parameters for which to judge performance might need to be identified early on by building management but the AI would eventually refine them over time based on what users interact with.

Another incredibly important feature of building management systems is anomaly detection. Natural language processing could interpret a nondescript alert and give it context. In a conversation with ChatGPT, ChatGPT explained, “a contextual awareness mechanism allows it to maintain a conversation state and keep track of previous interactions. This allows it to provide more personalized and relevant responses to users.” What’s that look like in action? If a building could talk it could say something like: “The second water meter is showing inexplicably high water usage; check for possible leak.” These alerts could even come in the form of an email or text message but in order to make sure that they don’t get overbearing, the AI would learn over time which anomalies need to be reported. A high water usage alert might not be tripped when cleaning crews are scheduled to work, for example.

The ability for the building to be able to answer questions can be monitored as well. If there were blind spots in the way data is being collected, the building could also report on that. This could come in the form of a statement like, “Electric meter three is experiencing a data outage, this is the third data outage in the past 45 days, recommend possible replacement.”

“Actionable building insights make building data useful for real estate operators,” said Gary Chance, VP of Marketing and Partnerships at Prescriptive Data. Their AI-powered tool Nantum OS has the ability to communicate what is happening in a building with managers in real time. Rather than just looking at historical data it can use its AI to make predictions. “Today, our software helps building engineers plan for tomorrow, recommending when to preheat their steam, suggesting the most energy-efficient BMS morning startup time, and even noting the probability of setting a new, costly, peak demand,” he said. “Many PropTech companies have been focused on historical reporting and real-time alerting for years now, but prediction and recommendations are at the heart of true building insights, and the precursor to automation.”

This type of self-aware introspection that AI can accomplish can help find correlations that building managers felt were missing from the first iterations of BMS software. By finding important correlations and explaining what they might mean, buildings are now able to direct managers’ attention to where they might not have been looking before. “Here’s an example of AI correlation: it rained for 106 hours last month, during those rainy hours energy consumption increased by 12 percent and interior space temperature increased by 2 degrees on the 3rd floor. Is this tenant uncomfortable during the rain?” Chance said. “Correlating tenant submeter data and temperature data, with weather and occupancy to develop a tenant retention insight is just one example of how NLG will change real estate management and create a competitive advantage for real estate owners.”

Right now, to understand what a building needs requires experts to spend a lot of time digging into the data. AI and natural language processing is helping make that process much easier. What might be even more important is that the same technology is also able to ask questions that a building’s owners and operators might not have thought to ask yet. We have the technology to not just deliver insights about buildings but also to put them in context and in a language everyone can understand. Now, there is no need to dig through mountains of data to understand what a building needs; they can just speak up and tell us.