In 2023, I wrote about a gold rush. At the time, a flood of AI-powered interior design apps were coming online, racing to reimagine America’s uninspired living rooms. There was some variety—some were scrappy operations, one was launched by Venus Williams—but the basic idea was the same: Put generative AI to work to deliver interior design to consumers on the cheap. Two years later, and the race is over. None of them won.
To be clear, most of these companies are still in business—and even now, there are new players entering the space. No one who Googles “AI interior design” will be hard up for options. And there have been some small improvements. These tools once subtracted and added windows and doors from a room at random—something that happens less often when using them today. Otherwise, most of the platforms have not managed to significantly surpass the capabilities they had at launch. Users still upload a picture of their space, click a few dubiously named style options (think: “ocean-inspired” or “professional”), and get an AI-generated image of OK-ish quality. The end.
While these tools have stagnated, another AI-powered app has gotten a lot better at interior design: ChatGPT.
To state the obvious, OpenAI, the company behind ChatGPT, did not build its product to compete with interior designers. Much of the early press coverage around its blockbuster 2022 launch focused on the chatbot’s ability to write emails and help college students cheat on English papers. But from the beginning, ChatGPT was capable of some basic interior design tasks. If you told the chatbot you wanted a small space to feel bigger, it would suggest putting up a mirror; if you asked it for tips on chalky, designer-favored paint, it would recommend Farrow & Ball.
This was all neat, but it was only text. Then, OpenAI introduced an update in 2023 that made the chatbot “multimodal,” meaning users could also upload image prompts for the tool to interpret. Subsequent updates to ChatGPT have made it more and more nimble with images, significantly improving its usefulness as a design tool.
You can now upload a picture of a drab-looking room and ask ChatGPT for advice. I wouldn’t put it toe to toe with Bunny Williams, but the responses are fairly good. (For example, ChatGPT told me, accurately, that the color palette of my living room was too dark.) The chatbot can also now “show” you what it means, generating renderings and moodboards based on your conversation. If you want real-world product recommendations, it can do that too—with OpenAI’s new shopping functionality, which debuted a few months ago.
In packaging all of these functions together into a simple chat interface, ChatGPT is significantly easier to use than most of the tools on the market that are specifically built for interior design. Users are taking notice: You can find Facebook conversations and Reddit threads where commenters go back and forth about how to refine prompts to get better results. (Clearly articulating your desired style, or using phrases like “keep the stove and sink where they are,” tend to yield more useful results; including your project budget can help field answers that fit your price point.) In April, Elle Decor surveyed the AI design tool landscape and found that ChatGPT was the clear winner, offering “something closer to an actual design consultation.”
ChatGPT is not the only chatbot on the market capable of these tasks. Google’s Gemini and Anthropic’s Claude are capable of similar feats. It’s also worth noting the usual caveats: All of these chatbots will get things wrong—notably, imagined details that the tech world has dubbed “hallucinations.” In my own experiments with Gemini and ChatGPT, they sometimes recommend products that do not exist, or slowly lose the architectural envelope of a room as the conversation progresses. These are real flaws.
Also important: The design advice, while solid, never felt inspired. These tools are built by hoovering up vast quantities of content and generating something like a mathematical average of it all. They are, by definition, generic.
But to put too much weight on these drawbacks is to overlook the very real progress these tools have made in a short span of time, and how much use they can offer the average consumer. That’s especially true compared to many of the early AI-powered design startups, which feel inert by comparison—it’s hard to shake the feeling that the only reason a homeowner would use one of them for design work is because they haven’t heard of ChatGPT.
When you zoom out a little, there’s nothing shocking about this dynamic. Some of the startups that launched in 2023 had a little bit of money to spend, but most were low-budget operations stringing together an app on the cheap. By contrast, OpenAI just raised $8.3 billion in its latest funding round. Google has said it will spend $75 billion on AI this year alone. The sheer amount of money these companies are pouring into developing better, faster, stronger AI tools is staggering.
However, there has always been optimism that smaller companies would carve out a niche and compete. The thesis, in a nutshell: If you make a tool specifically for interior design, customers will find it more useful than a general-purpose chatbot like ChatGPT. It’s early, and someone may still crack the nut and build a killer AI-powered design app, but a survey of the startups that launched in 2023 suggests that it hasn’t happened yet.
Indeed, perhaps sensing that they can’t compete on technology, many AI design startups have pivoted. Some now appear to be lightly scammy cash grabs targeting less tech-experienced consumers, sometimes inviting users to sign up for a monthly subscription before ever generating a single image. (As a rule: Free trials are industry standard among consumer tech products; requiring user payment before offering a test run can be a red flag of a less-legitimate operation.)
A more scrupulous move seems to be hunting for new customers: Many members of the Class of 2023 have shifted away from consumers and are now going B2B. Collov AI and Venus Williams’s Palazzo, are now marketing to real estate agents and brands; while the newer design-world launches like DecorX and Presti AI are specifically focused on furniture manufacturers and retailers.
There’s a logic to that. For individual consumers seeking a little advice about their living room, it’s hard to beat ChatGPT. But real estate agents, who may need to virtually stage dozens of properties a week, have different needs. Same goes for retail companies, which might want to “photograph” thousands of their products in AI-generated “rooms” in one fell swoop. Startups can offer features like batch uploading, project organization folders, and image generation presets that add genuine value. B2B seems to be much more fertile ground for these startups than DTC.
Whether this development should make (human) designers cheer is a complicated question. Early on in the AI boom, there was understandable fear that the technology was coming for designers’ jobs. More recently, the focus has shifted toward how designers can use it themselves—a recent Houzz study found that roughly a third of the industry is using AI in some form.
Anecdotal evidence would back that up. These days, I hear plenty from designers who are experimenting with ChatGPT, Gemini and Claude in new ways to help run a more efficient business. I have yet to hear from a designer who thinks they lost a real client to AI. The litmus test—can it get the painter to show up on time?—still applies.
But in small ways, anxiety rears its head. In the past, designers might have complained about a know-it-all client by saying something like, “They think they’re a designer because they spend all day tinkering on Pinterest.” Now, here and there, it’s ChatGPT instead of Pinterest. Of course, you can’t have a conversation with Pinterest.
Despite the firehose of AI hype coming out of Silicon Valley, it seems unlikely—to me—that what ChatGPT can do today is on the verge of costing designers a ton of work. After all, inspiration and advice has always been the easy part. If all the job took was a few tips and a list of products, designers would have gone extinct when Architectural Digest first started publishing sources in its captions.
On the near horizon, there is an iteration of artificial intelligence called “agentic AI” that promises to do much more than a chatbot can. Silicon Valley tells us that AI “agents” will take a wide variety of independent actions on our behalf—instead of simply recommending a sofa, an agent would be able to navigate to a retailer’s website, set up an account, and purchase the item with your credit card. OpenAI has already unveiled some agentic tools in its latest update to ChatGPT. They’re clunky, and they can’t buy a bedroom set just yet, but it won’t be long before they can.
The improvement of agentic technology will invariably change the business landscape yet again, and some potential clients may eventually choose to cobble together AI tools rather than hire a designer. But even in that version of the future, there is so much more to successfully executing a project than navigating the internet and clicking the right buttons—whether that’s managing a jobsite or finessing a complicated client or vendor relationship. Interior design remains firmly rooted in the messy complexity of the real world; the same thing that makes the profession so challenging is what protects it from the robots.
However, designers would be wise to take a real look at what tools like ChatGPT are capable of, not just to make their own businesses better, but to understand how they will shape clients’ expectations. Chatbots can already give genuinely helpful design guidance. They can generate decent renderings quickly. They can offer moodboards and product recommendations. And soon, they’ll be able to do some shopping themselves. If your business is built around doing these things, and only these things, it’s time to take notice.