I Touched AI — It Didn’t Bite

As it turns out, touching AI looked a lot more practical than you might expect. There were no robots. No machines plotting to take over the service department. Instead, I found myself standing in front of a touchscreen, looking at a simulated commercial building. 

PREDICTIVE: A demo of the BrainBox scheduling algorithm Khronos, which uses AI temperature prediction to optimize start/stop and reduce HVAC runtime. 

I tapped a button labeled "Preventive Maintenance." Within seconds, the system identified an underperforming rooftop unit, suggested several possible causes, and recommended a course of action. A few more prompts later, it had walked me through a diagnosis and generated a work order. This was a demo of BrainBox’s AI agent “Aria,” and it was pretty user-friendly. What would normally require digging through building data, equipment histories, and service records had been condensed into a simple conversation. 

As I watched the demonstration, I found myself thinking less about artificial intelligence and more about the HVAC professionals who will be using it — because whenever AI comes up, the first question we all ask ourselves usually isn't, "How much energy can it save?” It's, "Will this take my job?" 

AI’s Real Job: Eliminating Guesswork 

The executives I spoke with were remarkably consistent on this point. They all said AI is replacing something, but it’s not technicians. It’s guesswork. 

Anyone who has worked in commercial HVAC knows that fixing equipment is only part of the job. Before the troubleshooting even begins, there's a scavenger hunt. What BMS does the building use? Who has access to it? What equipment is actually installed? When was it last serviced? Has this problem happened before? Is there a manual somewhere? 

By the time a technician gets oriented and gathers all that information, a significant chunk of the service call may already be gone. 

“It would take two, three hours to actually get those answers,” said Riaz Raihan, Trane’s senior vice president and chief digital officer. 

That's one reason BrainBox and Trane executives repeatedly emphasized productivity rather than labor reduction. Dave Regnery, CEO of Trane, put it plainly: The goal isn't fewer technicians. The goal is helping technicians become more effective in front of customers.  

And finding information is only part of the challenge. Jean-Simon Venne, founder and president of BrainBox, pointed out that technicians are often dispatched with only the vaguest description of the problem: "It's not working." That could mean almost anything. 

Was this a one-time event or part of a recurring pattern? Has the issue appeared only under certain conditions? Has the equipment been drifting toward failure for weeks? The challenge here usually isn't a lack of skill. It's a lack of context, and without that information, technicians are forced to work backward from a symptom and piece together the story themselves. 

What AI promises to do, the execs told me, is provide more of that story upfront. By analyzing operating history, equipment data, and building conditions, AI systems can pull relevant information and narrow down likely causes before a technician begins troubleshooting. The technician still has to verify the issue, and they still have to make the repair. But instead of starting with a blank sheet of paper, they start with a list of informed possibilities. 

A Smarter Technician Is Still a Technician 

This proposition is much different than replacing technicians. In fact, the opposite argument was being made: It’s about helping them spend less time searching for answers and more time actually fixing problems — the thing techs like to do. 

In that sense, AI may end up functioning less like a replacement and more like a second set of eyes. Or maybe a very capable intern. 

Venne suggested that technicians should view AI as a tool that helps them solve problems faster and with greater confidence. I suspect that's the part that may resonate most with contractors. At the end of the day, customers don't care whether a diagnosis came from a service manual, years of experience, or an AI-powered building platform. They care whether the problem gets fixed correctly the first time. 

Don't Worry, Nobody Is Asking You to Learn Java 

It all seemed pretty straightforward to me, but one thing was nagging in my mind: A lot of HVAC professionals didn't get into this trade so they could spend their days talking to software. They got into it because they like solving problems, working with equipment, and fixing things. When conversations turn to artificial intelligence, cloud platforms, and digital agents, it's understandable that people can get nervous. For technicians who have spent decades turning wrenches, it can sound like they’re suddenly being asked to become IT specialists. 

So I asked the execs about that directly. 

Regnery acknowledged the concern, but argued that the technology is intentionally being built to be intuitive. In describing Trane's AI training efforts, he pointed out that technicians aren't being asked to learn a programming language or become software developers. The goal is simply to make information easier to access and use. 

“It's about making their lives easier and helping them do more,” Riaz added. “It is not about anything else.” 

That was one of the things that struck me during the showroom demonstrations. No one was writing code or trying to make sense of pages and pages of data. AI “agents” took care of that stuff behind the scenes. On the front end, most of the interaction looked a lot like having a conversation. Info was presented with graphs, recommendations, and easy-to-read analysis. 

In fact, one of my favorite moments came when I mentioned that many ACHR NEWS readers have spent 20, 30 years installing and servicing HVAC equipment and may be uneasy about all the talk of AI. 

Regnery's response was immediate. 

"They still are going to have to do that because it's a mechanical product," he said. "We're just going to tell them which wrench to turn at the right time."  

That may be the most practical description of AI in HVAC I've heard so far.