BrainBox Expands the Range of Facilities Management Solutions with the Addition of AI Virtual Advisor.

Siri, Alexa, and Google Assistant are out of date. To fully appreciate the new digital assistant in town, you must be a facilities manager.

The creator of an AI-powered facilities management system, BrainBox AI, unveiled ARIA (Artificial Responsive Intelligent Helper), a virtual building helper, on Tuesday.

According to the Montreal-based startup, ARIA is powered by AWS Bedrock and is intended to improve building efficiency by smoothly integrating it into daily building management procedures.

It was mentioned that ARIA, designed for business and retail settings, has the predictive capacity to avoid operational issues while keeping a close check on a building’s blind spots.

“We integrated an autonomous artificial intelligence agent into our pre-existing technology stack, which possessed predictive abilities,” stated Jean-Simon Venne, co-founder and chief technology officer of BrainBox AI.

“Consequently, it will not only provide you with an overview of the past year’s events, as well as an update on current events and upcoming events in the coming months,” he informed TechNewsWorld. “After that, it will offer some suggestions on how you should prepare for that immediate future.”

“It functions as a kind of ultimate advisor, advising you on how to handle upcoming situations,” he continued.

Encouragement

According to BrainBox, ARIA and its core AI for HVAC technology can have a big influence on building operations management by cutting greenhouse gas emissions and HVAC energy expenditures by up to 40% and 25%, respectively.

Facility managers’ morale may be impacted by ARIA as well. According to Venne, “building and facility managers are always frustrated because they have such a long list of to-dos.” They only have time to complete 20% of their list by the time their shift ends. They believe that because there are too many things to accomplish, they are failing.

“You have an assistant with ARIA that helps you complete a lot of the work faster,” he stated. “Now that you have a chance, you can finish a significant portion of that to-do list; previously, it was quite impossible.”

According to BrainBox, ARIA is intended to provide facility managers with a comprehensive overview of a building’s data, allowing them to focus on its systems and components to provide precise and well-rounded recommendations for strategic decision-making.

The business added that two-way interaction is deeply ingrained in ARIA’s architecture. Building operators and facility managers are not only given particular jobs to complete but they are also kept informed about upcoming measures that could result in the most effective and efficient administration of their buildings.

The business stated that customers could “call” on ARIA via text or voice and that their interactions would flow effortlessly from desktop to mobile.

Additionally, it stated that ARIA’s generative AI engine operates around the clock to assist clients in prioritizing and optimizing their facilities. This feature changes building management from reactive to proactive, raises the facility’s value, and directly supports an organization’s sustainability initiatives.

The use cases of ARIA are shown in the video that follows, showcasing how it may boost productivity, lower energy expenses, and streamline operational procedures in retail and commercial settings.

Preventing AI Delusions
One common worry with generative AI tools is their propensity to “hallucinate,” or generate responses to questions that appear plausible but are inexact or even bizarre. That behavior can be attributed to several factors.

Large language models (LLMs), for instance, are adept at imitating linguistic patterns but fall short of fully comprehending the meaning of the material they analyze. Because of this restriction, they might produce language that is grammatically sound but illogical or inaccurate in terms of facts.

An LLM’s decision-making process on the next words to generate might likewise affect hallucinations. Certain methods place more emphasis on fluency than precision, which can produce imaginative but unrealistic results.

Furthermore, a significant amount of data that has been scraped from the internet is used to train some LLMs. Factual errors, biases, and just plain strange things can be present in the data. Even if the outputs are inaccurate, the model can recognize these patterns and produce outputs that correspond to them.

BrainBox limits what and how ARIA uses data, so avoiding the hallucination issue. “Gen AI is a bucket of empty water,” Venne said. It must be connected to a data sandbox. After that, it might produce something intriguing.

“We’re integrating it with our current technology stack, which contains HVAC and additional building-related data,” he stated. “We constructed ARIA on top of the stack so that it can analyze data trends and have access to that vast amount of information.”

He went on, “Our data set is the only one in the AI sandbox.” The answer to a question like “Why did Napoleon lose the Battle of Waterloo?” might be something like “Can we talk about your building portfolio?” I’m not a chronicler.

Constructed upon Bedrock

According to Venne, one of the most difficult issues BrainBox has encountered in creating its AI-powered solution during the past six years is data translation.

“You rarely get a perfect data set when you plug yourself into these systems,” he remarked. Sensors can provide inaccurate readings. Gaps can occur, such as when a thermostat stops providing measurements for an hour. As a result, we needed to be able to extract accurate data and fix inaccurate data.

The speaker went on, “Using tools like generative AI is much easier once you have a super-structured set of clean data.” “Only data that we have complete control over was used to train our AI.

We have verified it and are certain of its accuracy.

AWS Bedrock is one of ARIA’s main components. “Bedrock allows us to maintain our flexibility without being locked into a technology choice that may not work out in six months,” Venne said.

According to Howard Wright, vice president and global head of startups at AWS, “Amazon Bedrock makes it easier for applications to leverage high-performing foundation models from leading AI companies like Anthropic, Meta, Mistral, and others, via a single API call — all in one secure, fully-managed service.”

“By leveraging insights from multiple models and selecting the ones that are most appropriate for a given use case or task, Amazon Bedrock empowers applications like ARIA to be the most sophisticated and intelligent building manager assistant it can be,” Wright told TechNewsWorld.

He clarified that by using foundation models, Amazon Bedrock eliminates the complexity involved in developing and growing generative AI applications.

“From Claude 3 to Llama 2, startups can quickly experiment with and evaluate top foundation models for specific use cases,” he said. “After that, they can create agents that carry out tasks utilizing the startup’s enterprise systems and data sources and privately customize them with the startup’s data using techniques like retrieval-augmented generation [RAG] and fine-tuning.”

“The challenge of heightening building sustainability isn’t just necessary—it’s urgent,” Wright continued. “Buildings account for nearly 40% of greenhouse gas emissions globally, with 27% of those emissions coming from the energy used to heat, cool, and power them. The global stock of buildings is expected to double by 2050.”

“BrainBox AI is rethinking energy systems and mitigating the effects of the climate crisis by using AI-driven technology trained on Amazon Bedrock to solve precisely this.”