AI advice for built-environment leaders
38thMove helps senior leaders in engineering, construction and infrastructure decide what to do about AI. The advice draws on 301 published AI examples, 53 surveys and more than thirty years in the sector.
Three recent stories from the evidence
15 hours → 4 hours
Research summaries
The engineering firm reports that Copilot cut a weekly research-summary process from 15 hours to four. The account does not say how much time went into checking and correcting the work.
Time saved on a useful task can create capacity. It is not a business-wide productivity gain.
Company account via Microsoft.
.
Read the case →
790+ hours
Construction document checks
Trunk Tools estimated more than 790 hours saved across four construction projects by helping staff check technical submissions. Staff still reviewed the findings and requested corrections.
Faster first checks are useful. The full cost and effort of getting work approved still matter.
Supplier estimate.
.
Read the case →
17.5% passed
Mechanical design test
In a test of 200 mechanical-design tasks, GPT-5.4 produced a file in 62.5% of tasks, but only 17.5% passed every required check. This was a software test, not a live building or infrastructure project.
Producing a file is not the same as producing work you can use.
Research paper, revised.
.
Read the case →
These accounts were published in the last 12 months; publication dates are shown above. Every example shows who reported it and links to its sources. See what the whole collection shows →
The evidence map
What we could talk about
There is no fixed package. Tell me what you would like to find out and we can work out together what would help.
An analysis of the database around your workflows, assets or market. It goes further than this site, with more detail and examples not published here. Where the evidence cannot answer your question, I will tell you so.
How the evidence analysis works →A conversation with your board or leadership team about what AI could change, what the evidence supports and which decisions to take next.
See the questions I ask boards →A second opinion before you invest: is the test fair, is the measure right, and what would it take to scale? Your team runs the work; you get an independent challenge.
See the methods behind the advice →Who you would be talking to
I’m an engineer by background, with early career experience in academic research. During 32 years at Mott MacDonald, I worked as an engineer before becoming IT Director and later Global Chief Digital Officer. I also created Mott MacDonald Digital Venture, a new region focused on innovation, digital technology and new business models.
I founded 38thMove to bring that experience to leaders working through what AI means for their organisations. It connects practical engineering, leading change in a global business and developing new ways of doing business. 38thMove is deliberately small: you work directly with me, drawing on that experience and the evidence to explore your questions and decide what to do next.
Why “38thMove”? In 2016, AlphaGo’s Move 37 looked wrong to expert observers, then proved brilliant and changed the game. 38thMove is about what happens next: how leaders respond once the game has changed.
Contact
You don’t need a brief or a decision already in mind. A short note about what interests you or what you would like to understand is enough. I will reply personally.
darren.russell@38thmove.com