Lately I've noticed something odd about my calendar. I'm in rooms that have almost nothing to do with each other: a client's fixed asset register, a working group of architects and building-data people, a podcast about consciousness, and late nights building my own software. And in every one of them, we end up having the same conversation. It's about AI, but it isn't really about AI. It's about trust, who gets to decide, and whether we can prove where our information came from.

The finance room

After thirty years and more than a hundred capital and construction-in-progress implementations, I can tell you that most fixed asset problems are not accounting problems. They're hand-off problems. Facilities replaces a rooftop unit and nobody tells accounting. The ledger shows a disposal while the insurance schedule still carries the asset. The class, life and cost get re-keyed by hand at the moment they're least verifiable, and the system then computes that upstream mistake faithfully for fifteen years.

The questions I ask in that room are simple. Where did this number come from? Who approved it? Is it still true today? And what can't we tell? That last one matters more than people expect. When no source can answer, the honest answer is “we can't tell,” not a confident “no.” In a real organization, that's the common case, and a system that hides it is not one you can trust.

I'm building those questions into my own software now, and the rules I keep coming back to are the same ones I'd want from any AI:

  • Every fact should say where it came from.
  • A finding that changes should be superseded, not erased.
  • Approving money is not the same as authority to act. A capital authorization releases funds; it doesn't give anyone, human or machine, permission to shut something off.

The building room

With Kimon Onuma FAIA, Michael Bordenaro and the Asset Leadership Network, the same questions show up wearing a hard hat. Kimon's phrase is “make intelligence more intelligent”: instead of having AI search disconnected information over and over, give it real relationships and context it can trust. Greggory Don Butler pushes on the next step, which is whether a proposed action has the evidence and authority behind it to actually happen. His shorthand is that capability is not authority. A sensor can detect a problem and an agent can recommend a fix, but neither one establishes the right to act.

What I find most encouraging there is that none of us had to give up our own system to make the connection work. Our separate tools can point to the same asset by the same identity without merging into one giant platform. Anyone who has lived through a failed attempt to put every department on one system will appreciate how sensible that is. The digital twin case study collects what that looked like when we tested it against one real building.

The podcast room

On my Love of One podcast, I've talked with Matthew James Bailey about AI twice now, once over a year ago and again this month in an episode that drops soon. Matthew comes at this from ethics, purpose and spirit. I come at it from the subledger. We keep landing in the same place anyway.

He describes AI as a thinking machine: it can reason and learn a field, but it's an intellect, not intelligence in the fuller sense. In my terms, AI can tell you the ledger and the insurance schedule disagree. It can't be accountable for what happens next. He also told business leaders to start with culture and then see where AI fits, and that one hit home. From inside the software world, I've watched companies move fast without a roadmap and let go of the very people whose knowledge makes AI useful. You can't prompt accurately for a capitalization decision after the person who understood capitalization has been shown the door. It's the old “the software ate my homework” excuse, now with a faster dog — and the same reason it pays not to automate a broken process.

The planet

The stakes in all of these rooms land in the same place. According to the UN Environment Programme’s Global Status Report for Buildings and Construction 2025–2026, buildings and construction are responsible for about 37 percent of global CO2 emissions and drive close to half of all material extraction worldwide, more than any other sector. A building is stored resources, and every asset on a register represents a decision to take something from the earth and put it to work. When the record is wrong, the waste is not only financial. A unit that was replaced but never retired keeps drawing insurance and property tax, and nobody can say what happened to the old one.

Matthew made a point when we were talking about smart cities that I’d apply to every building I’ve ever worked on: putting technology in to make something efficient doesn’t mean its resources are being managed well. The most consequential choices, about materials, repair versus replace, and even where a data center should go and whose water it will use, are made at the capital request, before the asset exists. That’s true for the books, and it’s true for the planet. The Design to Disposal lifecycle map shows where each of those decisions sits across an asset's life.

Where the rooms meet

So here is the conversation I keep having, whichever room I'm in:

  • Where did this information come from?
  • Who has the authority to act on it?
  • Is it still true today?
  • What can't we tell?
  • And when something goes wrong, does accountability stay with a person, or does it get handed off to the software, the spreadsheet, the staff and now the AI?

None of this makes me less enthusiastic about AI. I use it every day, and it has given me time back that I now spend painting and outdoors instead of in back-to-back meetings. It's a good partner precisely because I stay the one deciding. I used to keep the corporate side of my life and the conscious side in separate rooms. I don't anymore. One can be conscious and corporate, and the full life of an asset, from the first dollar requested to the last component retired, turns out to be a pretty good place to practice both.

Tomorrow I'll be in New York at the Greenbuild AI Summit with the Become the Building group, having this conversation one more time. I suspect it won't be the last.

If your organization is weighing where AI fits in its capital and fixed asset decisions — and who keeps the authority when it does — that is the kind of question Strategic Advisory is built for. Start a conversation →

The BIM project and Greenbuild

References