Everyone is talking about artificial intelligence right now, and it’s becoming a central topic in building systems. However, it raises an important question: If most building owners were not willing to pay for fault detection and diagnostics (FDD), why do we suddenly think they are ready to pay for AI? This question is not meant to dismiss AI, but to slow the conversation down enough to make it useful. 

There is no doubt that AI will play a central role in the future of buildings. However, when you look at how buildings are being designed, bid and delivered today, a noticeable gap exists between what AI needs to succeed and what most projects are actually set up to support.

And that gap did not start with AI.

FDD has been around long enough to prove its value. It offers better insight into system performance, earlier identification of issues and stronger long term efficiency.

Despite that, true FDD adoption has been limited. On many projects, it is one of the first things to get value engineered out. In many cases, even when it is included, it is not implemented to its full potential. Data is collected, but not always structured or used in a way that delivers the full value FDD is capable of.

Part of that comes down to cost. FDD requires additional eters, sensors and points. A single installed meter can approach $10,000. A chiller plant might need several. When those costs are added up across a facility, the number gets real very quickly.

However, cost is only part of the story. The larger issue is how the industry defines value during design and construction.

Still building for day one

Most projects are driven by a straightforward directive: get the building open on schedule, make it comfortable and keep it within budget.

“Efficient” often means the equipment itself is efficient: high-efficiency chillers, better air handlers, newer systems.

That all matters, but it tends to focus on what gets installed, not how the building will actually operate once it is occupied.

What gets much less attention is what happens after opening day. How does the building perform five years in? How are issues identified before they turn into failures? How do systems stay tuned as loads change and spaces evolve?

Those are day two questions, and the industry does not consistently make room for them. Owners are often removed from technical details. Engineers are under pressure to design quickly and economically. Construction teams are measured on execution and delivery.

So, the industry defaults to what it knows. Deliver 70-degree air, address the squeaky wheel and move on. 

AI runs into the same wall

This approach creates real friction when AI enters the conversation.

AI does not fix weak foundations; it depends on them.

If a building lacks proper metering, important information is missing. If sensors are limited, context is lost. If data is inconsistent or poorly structured, its usefulness drops off quickly.

Even something as simple as naming conventions can become a problem. The same point might be labeled differently across systems or even within the same building. People can work around it, but AI often struggles to.

Most projects do not spend the time or money required to clean it up. The reason is that they are not incentivized to do so.

Projects are still measured by how quickly they can be designed and built and how soon they can be turned over. In that environment, anything that requires additional coordination, standardization or long-term planning becomes harder to justify, even if it delivers value later.

As a result, AI is often layered onto buildings never designed to support it. Fixing it later will be a major cost and headache. In many cases, it means reworking the control system to create the foundation that should have been there from the start.

It is the same challenge FDD ran into. The technology was available, but the groundwork was rarely prioritized early enough.

Adoption is not only a technology issue

Even when the technology is in place, another challenge quickly arises: the personal side.

Take something like chiller optimization, where a system analyzes load profiles and determines which equipment should run for maximum efficiency. It might decide to run chiller two and chiller four instead of the sequence an operator has used for years.

From a data standpoint, it makes sense. From an operator’s standpoint, it does not feel right. That discomfort is real. 

It is the same reaction people have when they get into a self-driving car. Even if it is correct, it is not how they are used to doing it.

This hesitation slows adoption, even when the value is there.

A broader question the industry has not fully answered yet is who is actually in control. As more systems get “smarter,” there is a risk of having multiple platforms competing to optimize the same building. HVAC, lighting and security all operating with their own logic.

What you really want is one orchestrator. The building automation system should be the conductor. The chillers, air handlers and other equipment are only the instruments. If the conductor is not strong, or if multiple conductors try to lead at once, you are not going to get the performance you could.

A more grounded way forward

None of this suggests that AI does not belong in buildings. It does. However, the industry needs to first reframe its mindsets and approaches.

In many ways, FDD was the first real test of whether the industry was willing to invest in operational intelligence. In many cases, the answer was no. AI is the next version of that test, just with higher expectations.

A more practical approach is to be selective.

HVAC can account for roughly 30% of a building’s energy use. Within that, certain systems have an outsized impact on cost and reliability. Chiller plants, heating systems and other critical equipment are logical places to begin.

We should focus there first, show measurable results and build confidence before expanding further.

This approach aligns better with how owners actually make decisions.

Creating room for better decisions

This discussion is also an opportunity to rethink how time and effort are spent during design and construction.

A significant amount of energy still goes into resolving physical coordination challenges, such as routing ductwork and solving field issues. When that complexity is reduced, it creates space to think differently.

Prefabrication is one way to help. The industry already understands its value in terms of labor efficiency and schedule predictability. What is talked about less is how it can free up time to focus on controls, data and long term performance.

Instead of stopping at installation, teams can spend more time planning for service, maintenance and recommissioning. How do we keep the building tuned? How do we prevent drift over time?

That is where technologies like FDD and AI start to make sense.

The question beneath the question

So, the question remains. Owners didn’t pay for FDD. Why would they pay for AI?

The answer depends on whether the industry is willing to change how it thinks about value.

Are we willing to invest earlier in the infrastructure that supports long term performance? Are we willing to think beyond opening day and plan for how buildings operate years down the road? Are we willing to treat controls and data as core systems rather than optional add-ons?

If those answers do not change, AI will most often be discussed, piloted and eventually scaled back.

However, if the mindset shifts, AI becomes something more practical and achievable. Not a buzzword, but a natural extension of better planning, better execution and better building performance over time.

That is the real opportunity for the industry.

Diego Palacios, vice president of building automation national sales at Harris, has been leading high-performing teams to sales and business development success for almost two decades. His personable, consultative approach has helped clients and team members alike realize not only what today’s technology is capable of, but how it can address their needs.