BrainBox AI makes buildings greener and smarter with AWS Generative AI

Large commercial building managers these days have hundreds of thousands of data points to monitor and respond to in the ongoing care and maintenance of their facilities—including temperature readings, routine maintenance notifications, or alerts for sudden equipment breakdowns. A manager might sift through thousands of alerts per hour, perform root cause analyses, determine solutions, and send engineers or technicians to fix the problems. A single alert can lead to an hours-long diagnosis and generate redundant visits from contractors and engineers assigned to fix the issue.

Streamlining smart building maintenance

Jean-Simon Venne, cofounder and CTO of BrainBox AI, recognized that the slow but constant degradation that occurs in building operations overburdened building managers, and created increasing cost and complexity.

“As buildings essentially ‘untune’ themselves over time, it requires constant labor-intensive re-tuning by building engineers and technicians,” explains Venne. It was these pain points which inspired this Montreal-based company, BrainBox AI, to come up with an innovative solution, and with the added support from Caylent and Amazon Web Services (AWS), its generative AI-powered virtual building assistant ARIA (Artificial Responsive Intelligent Assistant) was created.

A leader in HVAC technology innovations, BrainBox AI is revolutionizing the way buildings consume energy. Since its launch in 2019, BrainBox AI has been steadily building the data infrastructure for autonomous building optimization, and the technology has impacted over 14,000 commercial buildings in over 20 countries worldwide. With the power of artificial intelligence, BrainBox AI has been developing solutions to make buildings smarter, while empowering building owners to reduce their carbon footprints.

Their latest innovation, ARIA, is a generative AI-powered virtual assistant that allows building operators and facility managers to interact with their structures in an entirely new way—conversationally and in real time. For instance, building managers can ask ARIA a question, such as “Can you compile a report showing the latest carbon emission trends from my building?” or receive alerts—in the form of a conversation—when an HVAC system is about to have an issue, then provide an optimized solution based on the specific part. At the heart of ARIA’s capabilities lies the Amazon Bedrock suite of foundation models (FMs), a powerful generative AI backend that enables ARIA to engage in natural, conversational interactions.

This emerging technology makes it so ARIA can understand complex queries, provide expert insights, and offer tailored recommendations for optimizing building operations. The ability to have this conversational back and forth is further enhanced by ARIA’s seamless integration with various building management systems, so managers can access real-time data and historical records, or engineers can get assistance at any time on the go.

And that’s the key advantage: ARIA is continuously learning and adapting to changing conditions. Through machine learning algorithms, ARIA refines its understanding of building dynamics, occupant behavior, and energy consumption patterns, enabling it to provide increasingly accurate and personalized recommendations over time.

The benefits of ARIA are multifold. When combined with their cornerstone AI for HVAC solution, buildings can operate far more efficiently, cutting energy consumption by up to 25%, reducing engineering labor drastically, improving tenant comfort, and most critically, cutting greenhouse gas emissions by up to 40%.

Generative AI has entered the building

Combining emerging technology of generative AI into HVAC and building data analytics and operations management was a multi-team effort. BrainBox AI leveraged Caylent’s innovative use of Amazon Bedrock and Anthropic’s state-of-the-art large language model (LLM), Claude, as well as other LLMs, as a conversational interface with generative AI capabilities and deployed ARIA on AWS for seamless data ingestion, analysis, and reporting. ARIA combines artificial intelligence, machine learning, and the latest generative AI foundation models (FMs) to provide “unparalleled efficiency and insight” to facilities and building managers around the globe.

“The beauty of Bedrock is it’s a carousel,” says Venne. “You have all of these models available. You basically flip from one model to the next. You could actually flip question for question. Or you could decompose a question, use a part of the question for Claude and use the other part with Meta Llama. And then you combine an answer from a Claude [model] and a Llama model, coming from Bedrock together, and you’re getting a very strong and very reliable answer.”

Traditionally, if (and when) something fails on a large commercial building, discovering the right solution to fix that root cause can take hours. With ARIA, the building manager has an AI expert at the ready that can quickly ingest all of the data and perform a root cause analysis. At a conversational level, ARIA can provide an alert as well as an insightful solution, such as: “The issue is in the HVAC system, and here are three possible solutions. Which one would you like to choose?”

“It’s more than just an LLM—it’s an autonomous AI agent that can operate independently within its specialized domain to execute complex procedures, update itself through feedback loops, and even converse naturally,” explains Venne.

With ARIA’s predictive capabilities, building managers can reduce the number of contractor site visits by providing the root cause up front, so contractors don’t need to figure out what part is required, then go pick up that part, and then return to the building to make the repair.

With current building management technology lacking the ability to provide predictive analytics and adaptive solutions to operate efficiently, it creates fatigue among industry professionals because they spend too much time responding to hot/cold calls in their buildings. On the flip side, ARIA tackles these challenges by taking in vast amounts of data, and then transforms it into actionable insights, elevating the decision-making process and offering a 360-degree view of building data for facilities management teams.

By changing the information exchange between building operators and their systems from a simple input/output relationship into a dynamic two-way dialogue, ARIA provides a scalable solution with capacity for autonomous action. As a result, this shift not only improves efficiency but also empowers facilities management with predictive and adaptive capabilities, reducing these hot/cold calls and significantly enhancing building performance.

ARIA, beyond buildings

ARIA’s unique ability to connect and train its generative AI capabilities with an entire body of building data so it can understand the intricacies of building operations means building managers can eliminate repetitive, time-consuming tasks. Those repetitive tasks, dealing with a building constantly untuning itself, means building managers can’t tackle bigger issues. And with more than 62% of building managers saying energy efficiency is a top priority for them, the ultimate vision of BrainBox AI is an ambitious one with major impacts.

According to BrainBox AI research, “HVAC systems account for around 35% of the energy consumed by retail spaces, office buildings, manufacturing facilities, and other commercial and industrial buildings.” ARIA was developed to serve as the essential companion to facility managers and building operators worldwide, helping them meet environmental challenges head on. And it opens up the potential for sustainability impacts at massive scale. BrainBox AI estimates that with the rapid pace of new building construction, equivalent to building a new Manhattan every month, ARIA’s generative AI capability in combination with AI-powered energy management could help deliver 25% in energy savings and contribute to a 40% reduction in building emissions.

“This is just the beginning—future iterations of ARIA could coordinate clusters of commercial buildings to actively reshape energy demand and prevent the spinning up of polluting power plants. We want to transform buildings from energy consumers into active participants and help reduce the effects of climate change. Through our partnership with AWS and utilizing Amazon Bedrock, we’re helping buildings reduce their carbon footprint and make a real difference,” states Venne.

The rapid and constant advances in FMs makes what seemed like the stuff of science fiction a few years ago possible today. “The ability to converse with an AI assistant like ARIA, it’s like Star Trek-level technology today. I never thought we would be able to see this,” says Venne. “Being able to stay on the cutting edge as the pace of technology continues to evolve makes working with AWS and Amazon Bedrock a gamechanger.”

Learn more:

Find out how you can transform your business with generative AI