Buildings have a big environmental problem. More than one-third of all greenhouse gas (GHG) emissions come from both the construction of new buildings and the heating and cooling of existing ones, according to the World Building Council. It’s why the United Nations Intergovernmental Panel on Climate Change has identified reducing building emissions as critical to meeting the goals of the 2016 Paris climate agreement.
When it comes to “green” buildings, most people think of the ultramodern, energy-efficient new construction that reaps architectural awards and LEED certifications. However, the real efficiency improvements are found somewhere less glamorous: the aging, workhorse commercial structures.
“Most of our buildings in Canada were built before we even had an energy code,” says Burak Gunay, an assistant professor of building science at Carleton University, whose work focuses on the use of big data and artificial intelligence (AI) in green building design. “There’s no easy way to ignore that if you want to really achieve the GHG targets we want to. New buildings are flashy and exciting, but the big gains will come from optimizing existing buildings.”
Optimizing the energy efficiency in older buildings is the mission of BrainBox AI, a Montreal-based company that has developed technology that connects to a building’s HVAC (heating, ventilation and air conditioning) system. The system uses AI, deep learning and cloud-based computing to pro-actively improve the energy consumption of buildings, dramatically reducing their carbon emissions and saving owners money. The system spends about six to eight weeks “learning” the daily rhythm of the building’s climate and occupancy patterns, based on hundreds of thousands of individual data points, right down to every bathroom fan or open window.
“When I saw the sheer impact of this technology, I just about fell off my chair,” says BrainBox AI president Sam Ramadori, a private-equity expert who joined the company in 2019, two years after it was founded, and has a long history of working in traditional heavy industries, including manufacturing and pulp and paper.
“In a pulp mill, you’re fighting to get efficiency improvements of like 2 per cent, 3 per cent. With this system, the technology achieves gains of 20, 30, 40 per cent,” Mr. Ramadori says.
BrainBox AI’s system is in use in more than 160 buildings in over 15 countries. Many of those efficiency gains are found here in Canada: For instance, a Quebec Holiday Inn has reduced its energy consumption in common areas by 34 per cent a month using BrainBox AI’s system. GWL Realty installed the system in a 300,000 square-foot office tower and a 500,000-square-foot residential building in Toronto, achieving efficiency gains of 29 and 25 per cent, respectively.
The technology was developed by Montreal-based entrepreneur and technologist Jean-Simon Venne, BrainBox AI’s co-founder, whose background includes maximizing the efficiency of buildings’ HVAC systems. The idea for the company came, in part, from an experience Mr. Venne had in the passenger seat of a self-driving car prototype in California several years ago. It was a transformative moment.
“He literally witnessed right in front of him this autonomous artificial intelligence working in real time,” Mr. Ramadori says. “It was moving him from A to B, working properly and not killing him.”
BrainBox AI applies a similar advanced AI to making buildings more energy efficient. Ironically, Mr. Venne’s technology is already in use, while the self-driving technology is not quite roadworthy.
BrainBox AI has received a handful of national and international accolades for its technology: It was named among the 100 “best inventions of 2020” by Time magazine, while the UN added the company to its “SME Climate Hub,” a listing of climate-friendly private companies contributing to the fight against climate change. Toronto’s MaRS Discovery District recently named the company one of its 10 climate champions.
BrainBox AI’s system works by marrying the data it collects with external information, such as incoming weather systems, and adjusts heating, cooling and other climate controls on the fly. Another smart feature: It also learns as the building’s conditions change.
An example is a building next to an empty lot: “Let’s say someone starts building a tower next door,” Mr. Ramadori explains. “If it’s to the south at the hottest time of day, suddenly that will cast a shadow on your building that wasn’t there before. The AI will learn that and see that the offices on the south side no longer need cooling [between] 10 [a.m.] and 2 [p.m.]. … That’s an insane level of learning no one would think to program into a building control system.”
And, once it has “learned” a building, Mr. Ramadori says the BrainBox AI system can continually predict climate conditions in each room of a building six hours in advance.
According to Carleton University’s Mr. Gunay, this ability to analyze occupancy patterns is another important aspect of smart building design, allowing for more fluid response than traditional preprogrammed building climate controls. It’s a design that could be increasingly important if the postpandemic world includes more flexible and less predictable work arrangements, as many expect.
BrainBox AI’s next step hinges on its recent hire, Nicolas Bossé, former manager of regulatory affairs with Hydro-Québec, as the company’s chief energy transition officer. His role will be to help integrate individual buildings plugged into BrainBox AI with larger power grids.
“We’ll take the 50 buildings in one area we may be plugged into and have those work together like an orchestra, to respond to the energy grid,” Mr. Ramadori says. “Imagine taking the 50 biggest in a downtown and saying, ‘Okay [everyone], the wind isn’t blowing much this afternoon, so the grid isn’t as clean, so for the next three hours we need you all to reduce your energy use as much as possible.’ This world is coming very rapidly, and we want to be ready for it.”
Transforming the proptech industry
Property technology, or proptech, is one of the fastest-growing areas for commercial AI. Here are three ways AI, machine learning and big data are transforming the industry:
Accurate valuation is the backbone of any real-estate transaction, but the number of variables involved can be dizzying; location, age, building condition and a host of external conditions.
A growing number of platforms are helping buyers and sellers analyze the enormous number of relevant data points to arrive at better valuations – so many that companies are now finding unique twists to distinguish themselves from the competition. Take OfficeBlocks, developed by global commercial real estate company JLL: Its Market Intelligence App can analyze a photo of a property, use a machine-learning algorithm taught by thousands of other photos, and generate estimated rent per square foot.
A suite of new platforms has come to market in recent years to help pinpoint emerging investment opportunities, highlight portfolio risks or even manage rental stock. New York-based Skyline AI, for example, analyzes local demographics and real estate transaction data to develop a prediction algorithm that forecasts trends and emerging investment opportunities.
More controversially, rental managers have begun using tenant-screening tools for residential real estate, analyzing everything from social media to legal history to algorithmically score prospective tenants – a practice that critics say risks entrenching existing biases.
Goodbye real estate agent?
Residential real estate is already well along the path to technological transformation. Online listings and sites such as Zoocasa have already made it easier for buyers to access market information without a realtor. And now AI is closing in.
Companies such as OpenDoor, known as iBuyers, use AI models to price residential properties for fast sales. Online real estate marketplace Zillow has recently gotten into this arena as well, and other companies are working on technology that can help buyers find properties as well as human realtors. In 2016, an experiment pitted real estate brokers in Colorado against an AI, to see which would find the best homes for clients. The robots won.