machine learning construction

How Deep Learning Is Redefining Building Management Systems

This Montreal-based company is pushing facilities managers to embrace artificial intelligence to optimize a building’s energy output

A Montreal company recently made Time Magazine’s Top 100 Best Inventions list for being the first in the world to use artificial intelligence-driven energy and comfort optimization technology for the built environment.

Launched in May 2019 by President Sam Ramadori and Chief Technology Officer Jean-Simon Venne, BrainBox AI uses self-adapting AI and advanced deep learning to transform the way buildings are heated and cooled. Inspired by self-driving technology that makes thousands of predictions every millisecond about the immediate future, Venne thought the same principles could be applied to HVAC automation to control temperature while also making buildings autonomous.

Buildings are among the largest climate change contributors, responsible for almost 40% of carbon dioxide emissions in the United States and 20% of greenhouse gas emissions globally. HVAC systems are responsible for a good chunk of those emissions.

Inefficient building systems are expensive for real estate operators to maintain, and there’s been little innovation in that space over the years, according to Jamie Hahn, BrainBox’s U.S. sales director. AI in real estate could be a game changer.

“Existing building management systems are very static, rule-based and schedule-based, but buildings are just the opposite: They’re dynamic and constantly changing—either with new use patterns, occupancy or weather—all impacting operations. Buildings need an adaptive solution,” Hahn said.

Communicating with the cloud in real time, BrainBox’s AI provides a building’s HVAC system with a peek into the future. Once the small black box is installed—which takes a couple of hours—it begins studying every zone of the building. Using multiple data points from the building’s access control systems, plus things like outdoor temperature, weather reports and utility information, BrainBox predicts a building’s thermal conditions, then identifies and implements all potential HVAC improvements. It keeps making micro-adjustments every five minutes.

“We’re able to proactively optimize around energy consumption while improving occupant comfort and reducing carbon footprint at the same time,” Hahn said.

Plug and play

BrainBox can be used anywhere there’s a centralized HVAC system that conditions and controls a large percentage of the square footage in a building, plus an existing building management system that supports open communication protocol.

“Unlike other solutions in the marketplace, we’re not here to rip and replace,” Hahn said. “We simply complement and improve upon the HVAC and control systems already in place.”

It takes six to eight weeks for the AI to form correlations between what’s happening inside the building and the impact of weather outdoors.

Using more than 25 customized algorithms working in real time, the BrainBox AI engine instructs the existing HVAC system to operate more intelligently and efficiently. As it gathers more data, it becomes smarter to predict how a building will react to future changes, enabling it to further increase energy and cost efficiency.

Standing out from the pack

With new technologies being introduced every day, BrainBox stands out because it not only collects analytics but also moves from reactive to pre-emptive operations management.

“Buildings are generating a tremendous amount of data each day,” Hahn said, “and there are products capturing that data and running some form of analysis against it. But someone—a human being—needs to do something with the insights in order to drive results.”

“We’re a closed loop-system, not only observing and learning, but implementing those optimization opportunities in real time,” Hahn added.

Once BrainBox learns a building, it can predict the future temperature in every zone of the building two to four hours ahead of time at an extremely high degree of accuracy—usually more than 95%.

“By knowing what the future holds, it pre-emptively makes optimal changes to the existing equipment’s strategy and settings, so we prevent or cancel out that unwanted thermal event before it ever happens,” Hahn said.

Proving its worth

To skeptics, a system that can generate significant energy and cost savings within three months sounds a bit like magical thinking. That’s one of the challenges BrainBox faces in getting the construction and facilities management industry to implement the technology.

“AI is fairly new to the real estate industry, so there’s a bit of engagement and education required on behalf of our clients,” Hahn said.

“It’s like an autonomous car: That first day you remove your hands from the steering wheel could be scary. But with what we’re doing, the building operator or manager doesn’t lose control of our system; they can grab the wheel back at any point,” he added.


Most energy efficiency building solutions also require large upfront capital expenditures, plus long timelines to deploy the technology. Because BrainBox installs quickly, it’s a very scalable technology, and its subscription-based pricing model means it can be deployed with little to no upfront capital expenditures.

The onboarding installation setup fee runs a couple cents per square foot, depending on the size of the building. Then, flat-fee pricing is based on the building’s square footage and energy spend, Hahn said.

“Typically, we’re seeing a net positive cashflow for our clients within the first couple of months. By the time we start optimizing the building, we’re generally providing enough savings that it offsets what our subscription cost is within that first two to three months,” he said.

The future looks bright

As capital market investors plan to mandate and deploy capital only into low- or net-neutral carbon assets, and as more major cities have adopted regulatory components and laws about decarbonization, tenants are demanding low-carbon spaces from landlords.

“With hefty penalties and fines coming down the road, I’m seeing a tectonic shift in the real estate industry to really look at decarbonizing their assets today,” Hahn said.

BrainBox technology can be found in more than 120 million square feet of building space—including hotels, office towers, grocery stores, airports and shopping malls—in 17 countries on five continents. The goal: Reduce those buildings’ carbon footprints by up to 40%.

Dane Ellerton, property manager of a 15-story office building in Montreal, Quebec, heard about BrainBox through a work associate and was intrigued enough to get the system installed soon after the company launched in 2019.

“We were looking to reduce our energy spend while also finding a better way to increase the comfort of our building for our tenants,” Ellerton said.

“The installation process was easy: We literally opened the door to the IT controls room in the basement, and the BrainBox AI team took it from there. We’re saving about 15% in energy costs per month, and one of the most positive things [that] has come from this process is the huge reduction in complaints from tenants: We used to get ten complaints for HVAC-related issues every week. Now we’re down to one or two.”

See more on how leaders can help their firms embrace evolving technology in construction.