Overheard at K12FF: What AI in School Facilities Actually Looks Like Right Now

At K12 Facilities Forum, we host roundtable discussions with district leaders responsible for facilities, operations, and capital planning.

These conversations are intentionally small so peers can compare how they’re navigating similar challenges across their districts.

They’ve always been closed-door and unrecorded until now.

With an AI note-taker and a promise of anonymity, we were able to capture the conversation without attaching comments to any individual district.

At the table, facilities leaders dug into a topic that is getting a lot of attention right now: how AI actually fits into school facilities and operations.

Not in theory. In practice.

The conversation moved quickly between data challenges, maintenance, capital planning, staffing, and the gap between what AI promises and what districts are actually able to use today.

Here’s what went down:

It Starts With Data.

The interest in AI is real. But the foundation isn’t.

Across the table, leaders described environments where facility data still lives in spreadsheets. Sometimes multiple versions of the same spreadsheet. Systems that don't connect. Information that has to be cleaned before it can even be used.

The challenge isn't understanding what AI could do.

It's getting the data into a condition where AI can do anything with it. For many districts, that work hasn't happened yet.

The First Use Case Is Efficiency

Where AI is showing up today, it's not making decisions.

It's helping teams move faster.

Leaders talked about using it to clean and organize data, speed up condition assessments, and simplify some of the most time-consuming parts of the job. One example came up repeatedly: equipment identification. Instead of a technician spending time tracking down manuals or specs, they can take a photo, and the system identifies the unit and pulls the relevant information almost instantly.

It's a small shift, but it removes a layer of friction that shows up everywhere.

That's where most of the value is right now.

Predictive Maintenance Is Closer Than You'd Think — In Some Places

There's a lot of interest in using AI to move from reactive to predictive maintenance.

Some districts are starting to explore this, especially where they have building automation systems and consistent data streams.

But most facility systems don't have that level of data.

Roofs came up as an example. Even with inspections, failures are still hard to predict in a meaningful way. The data isn't consistent enough, and some issues are inherently unpredictable.

A few leaders are already looking past predictive analytics toward what comes next: active monitoring through IoT. Real-time HVAC oversight. Transportation routing that adjusts on the fly. Food service systems that reduce waste based on actual usage patterns.

The idea of predictive maintenance is appealing.

The reality is still early. But the direction is starting to take shape.

Trust Has to Be Earned, Not Assumed

One leader shared a cautionary tale of an AI tool generating fabricated information for a cost estimate. Confident, well-formatted, and wrong.

That story landed because everyone in the room understood the implication.

If the AI is wrong and no one catches it, the consequences aren't just embarrassing. In capital planning, they're financial.

The conversation turned to how trust actually gets built. One concept that came up: a "confidence map." A way for the system to show how sure it is about a given piece of data, so reviewers know where to look closely.

AI doesn't earn trust by being right most of the time.

It earns trust by being honest about when it isn't.

AI Works With People, Not Instead of Them

No one in the room is handing over decision-making.

AI is being used as a support tool.

It can handle a large portion of the work, but it still needs to be checked, validated, and interpreted—especially when the stakes are high.

In capital planning, a bad output doesn't just create a small error. It can lead to real financial consequences.

Human oversight isn't optional.

The Hard Part Is Standardization

The biggest barrier isn't access to tools. It's consistency.

Facilities, capital planning, finance, and operations often all manage data differently. Different naming conventions. Different systems. Different levels of detail.

At one point, the conversation shifted to something more fundamental: language.

If different teams are describing the same asset in different ways, the data doesn't connect. And if the data doesn't connect, AI has very little to work with.

This isn't just a technology problem; it's an organizational one.

The Power Question on the Horizon

Toward the end of the conversation, someone raised a different kind of concern: power.

AI consumes significant computing resources, and the cost—both financial and environmental—isn't trivial. For facilities leaders who are literally responsible for energy budgets, the question landed differently than it would in most industries.

The response from those working in the space: the long-term strategy is to "learn once, cache forever." Use AI to build structured data libraries, then run cheaper lookups against them rather than running expensive models constantly.

Whether that strategy holds up at scale is still an open question.

But it was the first time in the conversation that the room felt like it was looking past the next twelve months.

Some Districts Are Moving Faster

A few districts are further along.

Using AI for administrative work, contract review, and supporting day-to-day tasks. Beginning to test how it might fit into operations.

Others are earlier.

Encouraging staff to experiment. Using basic tools. Trying to understand what's actually useful and what isn't.

The range is wide.

But the direction is consistent.

The Workforce Shift Is Already Happening

There was also a shift happening underneath the conversation.

Facilities teams are managing more data than ever before, while at the same time losing institutional knowledge as experienced staff retire.

That creates a gap.

Veteran technicians understand systems from years of experience. Younger staff are often more comfortable working with data, but don't always have that same field intuition yet.

AI is starting to show up as a translation layer between those two. Taking the patterns veterans recognize instinctively and surfacing them as data younger staff can interpret.

Not replacing expertise.

Bridging it.

The Work Right Now Is Getting Ready

The conversation didn’t land on a single solution.

It circled around a shared reality.

AI has real potential in facilities, especially in planning, data management, and maintenance, but most districts are still doing the foundational work required to use it well.

Right now, the focus isn’t on fully adopting AI. It’s about getting ready for it.