4 answers
★ Best answer
I've used it quite a bit with architects and the experience really depends on what you expect from it. What works well is that initial brainstorming phase - you generate three layout concepts in minutes and then keep refining them. But there are always some annoying issues: the AI tends to ignore local constraints (like ceiling height or existing windows), and then you spend time fixing what looked right at first glance. Basically it's a tool to speed up ideas, not to replace someone who actually understands the space.
I've never used AI for that, but from what I see people saying, it works well for generating initial ideas and exploring concepts quickly, but you end up spending time later refining and correcting because it lacks that human judgment about what actually makes sense for the project.
Don't fall into the trap of thinking AI will generate a finished project - in reality, it usually produces pretty generic stuff or makes proportion errors that you only notice after poking around a lot. The biggest issue is that you end up spending more time fixing structural details than you would if you'd built it from scratch, because AI gives you this false sense that it's almost done.
AI is way more capable now in 2026 than it was a few years back, but it still has real limitations that people don't always notice at first. What I see happening most is people jumping in thinking they'll save hours and then realizing it actually eats time because it needs a ton of refinement - you generate a layout, then you see the proportions don't match the actual square footage, or the circulation feels off, or it breaks local codes.
My experience with it is that it works better when you use it as a second opinion, not as a replacement. Like, you already have an idea of the solution (or at least the project constraints - lot size, zoning, what the client wants), and then you ask the AI to generate variations on that. That's when it actually saves time. But if you just throw "generate me a residential project" at it with no context, you're gonna get something way too generic, out of scale, or that misses details that matter.
The worst is trusting that first render too much and showing it to the client without a thorough review. AI loves making a visually stunning project but with construction errors, weird flows, or spatial relationships that don't actually work in practice. It's better as a tool for exploring ideas and quickly communicating concepts than as a final deliverable.
Your answer
Log into answer.