The July/August issue of AEC Magazine landed last week and, as always, is packed full of great content. (And even a couple of my pieces made this edition.)
While there are a lot of great articles, two articles in particular stood out to me and deserve reading.
The first is a piece by Martyn Day that takes ALPA's Architecture Studio — the same open-source repo I wrote about in my previous post — and builds a framework around what it actually represents.
He calls it the Judgement Stack.
The model is three layers:
- Vendor AI sits at the base: the assistants and copilots shipping inside Revit, Archicad, and every other major platform.
- Above that is the judgement stack itself — the skills and workflows a practice builds to encode how it actually works, not how a software vendor thinks all practices should work.
- On top of this is institutional memory, where those skills stop performing isolated tasks and start reasoning over decades of real projects.
Each layer builds on the one before it, but each also becomes harder to copy.
He argues that competitive advantage sits in the second and third layers, not the first. The first will become table stakes, the same way rendering and clash detection did. The second and third are specific to your practice. Nobody can ship them to you.
That is the opportunity argument.
The second article, Richard Harpham's piece later in the same issue, is the risk argument.
This isn't an argument against AI adoption. It's simply an argument that proposes that adoption without verification discipline is a liability problem wearing a productivity costume.
And:
The pace at which AEC's culture, contracts and ways of working evolve are radically out of sync with the pace at which the underlying technology is now moving. That is the actual risk to every technology strategy in this sector.
Read alongside each other, the two pieces are less competing arguments than two parts of the same one.
The Judgement Stack is not just the opportunity. It is also the governance mechanism. Codifying when to apply conservative professional interpretation, what checking means for a given output type, who signs off and on what — that is both the competitive asset and the accountability structure. You do not get to build one and defer the other.
Most firms are treating adoption and governance as sequential. Adopt first, sort out accountability later. What Harpham is describing — and what the pace of change makes harder to ignore every quarter — is what happens when later keeps being postponed.
(Two pieces of mine also made the July/August issue: NXT BLD 2026: One Direction of Travel and The Word "Agent" Is Doing Too Much Work.)