No one gets into architecture with a passion for code compliance. If you did, no offense, we want your warm and inclusive perspective. But for the rest of us, the code arrived in our lives as a threat.
The scare
I was in architecture school through the Great Recession, and our professional practice classes had a particular flavor: alumni coming back to warn us. You are going to have a bad time. You are going to do ADA diagrams and stair details. You will never design. It will crush your soul. We were all quite a ways down the line by then, so it was slightly too late to turn back, but the message landed. Code compliance was the looming spectre waiting on the other side of graduation, the thing that ruins your vision after you have drawn it.
I think that is an antiquated view, and I say that as someone who spent fifteen years in practice doing the diagrams and the stair details, and who comes to AI with a decent amount of skepticism and a healthy defensiveness of this profession. The spectre is real, but it is not the code. It is the way we encounter the code: late, manually, and after the design decisions are already made.
Is compliance a negotiation?
On a podcast recently, the host told me about a New York real estate attorney whose entire philosophy of code compliance came down to one line: everything is a negotiation. I understand where he got it. One of my first firms did several buildings for a client in the jewelry district, and we developed an interpretation of the zoning code where pulling the floor plate back about two feet turned the vertical face of the building into something you could sell as a billboard: not signage, so not controlled as signage, and not leasable interior space either. A whole new tenancy on the face of a building, resting entirely on an interpretation. Nowhere does the code say you can do that. Nowhere does it say you cannot. That is negotiation, and honestly, it is some of the most creative work I have ever been part of.
But nobody negotiates a turning radius. Nobody negotiates knee clearance, stair widths, or the fire rating between occupancies. A huge share of the code exists to guarantee quality of life, equity, and safety, and it is written to be binary on purpose. The model either meets the requirement or it does not. You do not want to negotiate on that, and neither does anyone who will ever use your building.
The 80/20
So both camps are right. The lawyer is right that interpretation is real, and the engineers are right that most checking is mechanical. The proportions are what matter: if 80% of compliance is black and white, then you provide those things, and the 20% that remains leaves you a wonderful array of levers to push and pull.
The black and white 80% is exactly the work that software should check continuously, because there is nothing interpretive about it and everything repetitive about it. The 20% is where your judgment, your reading of the jurisdiction, and your billboard-that-is-not-signage ideas live. There is an old Eames line that constraints are opportunities. It only works if you actually know the constraints, and know them early enough to wield them. If you know enough about the code constrictions early, it does not rob you of your design. It is not a roadblock. It is a tool.
Envelope first
The New York City planning guide hands you diagrams: the base, the setback, the sky exposure plane you need to clear. The lowest common denominator of compliance with those diagrams yields boring buildings, and the city is full of proof. But read them differently and they describe an envelope, an invisible fencing. Understand that fencing first and you can build anything you want inside it. The constraint stops being the thing that trims your design down and becomes the thing you sculpt against.
The same logic scales down to a single detail. I spent years in boutique hospitality, and the bar is always where design intent and accessibility meet. The ADA drop, the knee and toe clearance, the turning radius: all binary, all non-negotiable. The design question is whether you get to iterate. If checking your grab bars, fire separations, and door maneuvering clearances consumes four days of the team's week, you do one version of the sculpted stone bar and hope. If the checking is running against the model as you work, you do twenty sketches of it, every one of them compliant, and pick the most beautiful.
Completed thoughts, not popups
Real time does not mean naggy. Nobody wants to move a door and get a violation popup. The rhythm that actually fits how designers work is: I have completed this thought, let me make sure it works the way I think it does. You are shifting walls, playing with spatial layouts, weighing option A against option B, and you want data at the moment of comparison. This option gets us a few more bar seats but loses two banquettes. This one gets us two more floors, 30% smaller. Compliance is one more input in that trade, delivered while the trade is still live.
And there is a second effect that I care about even more. When a PM says give me four options for this classroom layout, a young designer can now show four options that already work, before the senior review, instead of learning by demoralizing redline. Because every finding traces to a model element and a code section, the code stops being someone senior being petty about your dreams. It has a source. You learn where it comes from. It does not take the 10,000 hours of studying door and wall ratings for compliance to become second nature; it becomes a reward cycle instead. That is not replacing junior architects. That is how we bring people up through a profession that is about to lose a generation of tacit code knowledge to retirement.
What do you want back?
The time you get back only matters if you are intentional about what it becomes. Productivity and creativity can be delineated from one another, and they should be: the point of getting faster at the checkable work is not to cram in more checking, it is to protect the part of the week where the thorny questions live. When I ask myself what I want back, the answers are embarrassingly analog. I loved hand drawing in school and stopped having time for it. I said on a Dezeen panel this summer that an architect will always be a storyteller, and stories have to come from somewhere other than a screen.
Some of that somewhere cannot be outsourced even in principle. Early in my career I learned to tell plumbing from gas from electrical conduit on a site walk, because they are three different metals and we draw them all the same. No tool teaches you that, and no tool needs to. Walking the site, looking at open walls, talking to the cabinet maker: that is the kind of time worth buying back with software, and it feeds directly back into the work.
Wield it
Architects are allowed to keep our identity of being somewhat frustrated. I certainly have. The trick is directing the frustration at more productive aims than checking sink clearances by hand. Handle the 80% that is black and white the moment it can be handled, and the 20% that requires a human reading stops feeling like a threat and starts feeling like the job: interpretation, judgment, and the occasional billboard where a wall used to be.
No one gets into architecture with a passion for code compliance. You do not need one. You need the black and white handled, and the good part of the job back.