How A.I. Will Change Architectural Clients in 2025
Design in ‘25: The Near Future of Architecture & A.I., Pt. 4
In This Post:
Unpacking 2024 . . .
In 2025 . . .
Architects Are in Their Client's Business, And That Business Is A.I.
Latent Threat of ‘Design Your Dream House’ A.I. Platforms Emerges
Municipalities Wander into A.I. Waters
‘Guardrail’ Practices Go Mainstream – You Buy One
Unpacking 2024 . . .
For 2024, I had had three predictions about clients, but one main one: that clients will start to ‘expect’ A.I. to show up in an architect’s work. How that A.I. shows up is of course something we’re all still feeling out, and that will continue. I predicted we’d be having a robust conversation about the surveillance state, and what role architects will play in it, but that hasn’t fully materialized. I also predicted that insurance markets would retreat due to climate change, and that is definitely happening.
But generally, we can expect this general trend of ‘expectations’ to rise. Clients are increasing their use of AI, and they’ll expect that to show up in an architect’s work accordingly. I’m watching three possible manifestations of this expectation in 2025:
In 2025 . . .
Architects Are in Their Client's Business, And That Business Is A.I.
As more and more of the general client body out there starts to use and understand A.I., that can't help but inform the client's priorities. Somewhere, an executive is considering building a new facility, and asking themselves some questions:
Is A.I. going to trim my workforce by half? If so, how many desks should I plan for in the new facility?
If my entire business model is moving to A.I., and that's in the cloud, how critical is the location for the new facility? Does it need to be where I am? Where my clients are? In the city center? Or on the periphery?
A mediocre architect will sit off to the side and wait for their would-be client to figure all these things out, and then present them with a program. A ‘good’ architect is invested in their clients and is the client’s partner in figuring out these questions. For that reason, architects need to be understanding not just how they’re going to use AI, but all the ways in which their clients will (or should) be using it. This may even provoke a shift in client bases. If a particular bread-and-butter client is being stubborn in their A.I. adoption, and their industry isn’t, you might consider the possibility that that client won’t be around much longer.
Latent Threat of ‘Design Your Dream House’ A.I. Platforms Emerges
The evolution of clients can and will extend to clients wanting to act as their own clients. I discussed this earlier under The WebMD-ification of Architecture, but want to specifically highlight the rise of A.I.-driven ‘design your own house’ platforms like Vitruvius.
Most architects I talked to this year were relatively unconcerned. Architects don't work in residential housing. And such programs haven’t led to an explosion of custom houses for the middle-class (they won’t). But that’s not really the point. These programs will scale.
Smaller companies, using smaller models, to design smaller buildings, is the proving ground in which these algorithms will be built, tested, and improved. Some of those companies will go bankrupt in the process. But the successful survivors will have a refined technology that is then more easily scaled up towards larger, commercial projects. Some startups already boast the ability to produce an entire set of CDs for a single house given only an address and client input. In 2025, look for signals that some of these players will move to scale up their technology.
Municipalities Wander into A.I. Waters
Also look for municipalities to scale up their tech, and for startups to target them as potential customers, as well.
Permit review is slow. As with many bureaucratic functions, it's slow because it doesn't need to be fast. So, we shouldn't expect municipalities to speed up (via A.I. or any other means) unless they have a reason to do so.
Municipalities face pressure to speed up plan review from developers, elected officials, housing advocacy organizations, etc. But there's no real way for them to do what, unless they hire more plan reviewers - and that takes appropriations, and blows up the budget, which then gets them in trouble from the other direction. Even then, cities like San Francisco haven’t hired enough people to hire people!
A.I. offers a third path. With comparably little spending, a planning department can deploy A.I. against plan review, and retain their existing reviewers to supervise and check the A.I.’s work. While certain companies like UpCodes have been tackling codes & code review from the designers perspective for a while, startups like Autoreview.AI are developing automated code review platforms specifically marketed to municipalities, and has found users in Gainesville, Pasco County and Altamonte Springs.
‘Guardrail’ Practices Go Mainstream – You Buy One
For clients, municipalities and architects alike, the full adoption of A.I. depends on risk mitigation. Any A.I. adoption needs to bring down the overall level of risk, and there’s no incentive for adopting A.I. if you’re just trading your traditional forms of risk for new, less-understood A.I. risks.
For that reason, we’ll see an expansion of the A.I. guardrail industry - companies, products and services specifically designed to mitigate A.I.-native risks, like hallucinations. If a company wants to fully adopt an A.I. process, but fears that a random one-in-a-million hallucination could ruin them, they can look to companies like Guardrail AI to design custom mitigation solutions.
The ‘guardrail’ industry is brand new, obviously, but look for it to grow in 2025. Also look for some startup to create AEC specific guardrail technologies and services.
If you enjoyed that, be sure to check out Part 5 of this series: 'How A.I. Will Change the Way We Feel About ‘Design’ in 2025’ and subscribe below for all future updates.