Residential construction is adopting AI faster than most people realize. Not because the industry suddenly changed, but because AI finally offers interfaces simple enough for contractors, developers, and homeowners to use. We spoke with Patrick Murphy, CEO and co-founder of Maket, about why adoption is accelerating, how the architect's role is evolving, what the approval process needs to change, and where the industry goes from here.

The construction industry has a reputation for being slow to adopt technology. The real story is different. It's not that construction professionals don't want better tools. It's that the tools they were offered for the past two decades were built for a completely different kind of user.

Patrick Murphy is the CEO and co-founder of Maket, an AI platform now used by over one million people to generate floor plans. His background isn't in architecture or construction. It's in growth, paid advertising, and product. That outside lens is exactly what makes his read on the industry worth paying attention to. He thinks less in terms of tradition and more in terms of adoption curves, user behavior, and where the data points.

The real adoption problem

The adoption problem in construction isn't about willingness. It's about design.

"Yesterday, a plumber came to my house to repair my toilet. He was saying, 'I need to log all this information in this application, and it's hard for me to understand how to use it. All these young guys can do it quickly.' I looked at what he was doing on his phone, and the interface was dead simple. A bunch of yes/no questions, a signature, and that was it."— Patrick Murphy, CEO & Co-founder, Maket

That's the whole problem in one scene. Construction technology has been designed around complexity: long onboarding, deep feature sets, enterprise sales cycles that take months to close. But the people who actually need these tools (on job sites, between client meetings, in the middle of a project) don't have time for any of that. They need to get to a decision and move on.

"We've been used to these enterprise systems that are long to integrate, long pilot projects. The reason we're not seeing a lot of these companies adopt AI is because they never get past the pilot phase. What we need is product-led growth style companies. Take a look at Notion. Take a look at ChatGPT. Claude. These companies have made using the systems really easy."

Why AI is unlocking what previous tools couldn't

Construction has had access to specialized software for years. CAD, BIM, Revit, SketchUp, various SaaS platforms. The tools exist. The problem is that adoption has remained low outside of large firms and specialized roles. For contractors, developers, and homeowners, the learning curve was too steep and the interfaces too far removed from how they actually work.

"These people are using Claude because they want answers to their questions quickly. They want to generate bills of materials. They want to do cost estimations, takeoffs. They're trying to generate floor plans using ChatGPT and Claude."

That's the shift. Not that previous tools didn't exist, but that a much wider group of people is now able to get useful output from technology for the first time. A general contractor doesn't need to learn Revit to explore a layout option. A homeowner doesn't need a SketchUp certification to test whether a garage fits the lot. They can design a house with AI from scratch. They describe what they want in plain language and get a result.

Looking ahead, there's a second layer to this that goes beyond interface. AI systems are increasingly able to read and interpret different data formats, which opens the door to connecting information that has traditionally lived in separate, incompatible systems (zoning data, design files, cost estimates, building codes). As generative AI matures, the ability to bring all of that into a single conversation is what will make it fundamentally different from the tools that came before it.

What one million users revealed

Maket went from zero to over one million users in under three years. That kind of growth teaches you things that market research never will. And for Patrick, most of those lessons were surprises.

"One of the things we really got wrong in the early days is we thought it was gonna be by architects for architects. One of my cofounders comes from the architecture industry. We said, 'How do we make your past life easier?' We ended up being super surprised. The actual people using it are consumers, real estate developers, homebuilders. Not the architect."

Architects already have deep, specialized tools woven into their workflow. The unmet need wasn't there. It was everywhere else: homeowners planning a build, developers exploring site options, contractors who need to show a client three layout variations before the next meeting. A massive market that had no tool at all.

The other surprise was specificity. People don't ask for "a 3-bedroom house." They describe walk-in corridors to saunas, parking garages in unusual shapes, kitchens oriented toward specific views. Every project has its own long-tail requirements. No template library can handle that. But a text-based AI system can, because it adapts to the prompt instead of forcing the user into a fixed set of options.

"I've also been surprised at how basically simple the core feature set needs to be. It's plan generation. It's editing. It's visualizing the space. It's bringing in your own existing space to make changes using a text-based approach. That's really what makes us unique."

That's the product insight that drives everything at Maket: people want to describe what they're imagining in their own words and see it become a floor plan. Not learn software. Not sit through onboarding. Structured input for the basics (floors, square footage, shape), then text-based prompts for everything specific to their project.

The architect isn't going away. The process is.

If AI is handling the early design work, what's left for architects? Patrick frames it through a lens that might surprise people in the industry, but makes perfect sense if you've been watching what's happening in tech.

"Architects will go from more 'doer,' in terms of working in CAD software and legacy systems, to 'tastemaker' and approver of what's been done by generative systems. A lot like what we're seeing with software engineers. There are a lot of AIs writing the code, and what engineers are doing is making sure the code is actually functional, works, and integrates within a more complex structure."

This isn't theory for him. When renovating his own kitchen, Patrick used Maket's technology to explore layouts and options. He went through the full process. And at the end, he still wanted a professional's eye on it.

"We ended up meeting a company that did our kitchen cabinets. We got a final approval by somebody who's expert in kitchen design. I see the architect in the same fashion. They're always gonna be giving that final stamp of approval, that final taste approval, on what the AI is generating."

For residential, that means architects shift toward guiding and validating rather than producing every iteration from scratch. (For a deeper look at how this plays out in practice, read our conversation with Maket's co-founder, Stéphane Turbide.) For commercial and institutional projects, the picture is different entirely. Those involve magnitudes more complexity in documentation and zoning codes. Generative systems aren't there yet, and architects are still doing the heavy lifting. That's not changing anytime soon.

How AI can help address the housing crisis

The housing crisis has many causes: material costs, interest rates, labor shortages, land scarcity. But there's one that gets less attention. The process itself.

"The way that we build buildings and approve building permits has basically remained the same for the last fifty to a hundred years. Generative AI is the big unlock, because you have these systems that can now speak to each other. Different data formats can be read by different AI models, and they can start talking to each other."

Consider what happens today when someone wants to convert a commercial space to residential, or rezone a property. The design goes through an architect, gets submitted to the city, comes back with changes, goes through another iteration, gets resubmitted. Each cycle adds weeks or months. And the outcome is uncertain because there's no reliable way to know in advance what will get approved and what won't.

That's what Patrick sees changing. As AI systems get better at interpreting zoning requirements, building codes, and compliance constraints, they'll be able to produce designs that fit within those rules from the start. We're not fully there yet, but the trajectory is clear. And when it happens, the value won't just be faster design. It'll be compressing the entire approval pipeline. Fewer rounds of back and forth between designers, builders, and municipal authorities. That compression is what will actually move the needle on cost. Not cheaper software. Faster decisions.

Where it goes from here

Patrick thinks the next big shift isn't in the models. It's in the interface. Text-based interaction with AI is already transforming how people design homes. But text is just a bridge.

"Even the way I've used it has changed. I used to look for a specific output, put in a text prompt, get the answer, and leave. Now I'll have 45-minute conversations with Claude using voice, back and forth. That's gonna make its way to how everybody uses generative systems. Text is just a necessary step on the way to voice."

Conclusion

Residential construction didn't adopt SaaS the way other industries did. And that might actually be an advantage. Instead of being locked into legacy software with years of sunk cost, the industry is moving directly into AI with a clean slate. The interfaces are simpler, the output is faster, and the barriers to entry are lower than anything that came before.

One million users have already proven the demand is real. People are designing homes with AI today, not because it's trendy, but because it solves a concrete problem: getting from an idea in your head to a visual plan you can act on, without months of waiting or thousands of dollars in preliminary design fees.

The construction industry wasn't slow to adopt technology. It was waiting for technology that actually fit how it works. That technology is here.