AI Interior Design Gold Rush Now Ongoing

Now somewhere in a far away data center, a computer describes a room. We use cutting-edge machine learning techniques, algorithmic magic, and computer vision to delve into the contents of a modern farmhouse kitchen and catalog its elements. tables, chairs, subway tiles.

This process is part of an artificial intelligence training program devised by a startup called Reimagine Home. Their hope is that by letting AI understand room content in natural language, it can better understand what humans really mean when they say, “I want a modern farmhouse kitchen.”

It’s trippy. But these are trippy times. As artificial intelligence engines such as GPT-3 and Stable Diffusion moved out of the lab and into general circulation last year, a frenzy began to turn them into consumer products. Some, like the AI-powered selfie tweak tool Lensa, have become viral hits on social media. Others create marketing copy and fight parking tickets. Of course, some people do interior design.

The first platform to catch the public eye was Interior AI, which debuted last October. At the time, its designer and developer, Peter Revels, said: Residential business His creation is just the beginning.[Image-generating AI models] Stable Diffusion, Dall-E, etc. are only a few weeks or months old. Very early. We are imagining more products soon. ”

The level was correct. A mini boom in AI-powered interior design apps is already underway. Some are derived from existing companies, such as Reimagine Home and CollovGPT. Others like RoomGPT were built by independent developers. There are serious endeavors for dedicated founders, and knowledgeable side hustles for entrepreneurs who are still running their day jobs.

chance
Regardless of size or purpose, these apps definitely have an audience. Users flock to her AI-designed site. RoomGPT has had 3 million visitors to its site, says founder Hassan El Mghari. Meanwhile, founder Akhilesh Majumdar says Reimagine Home has hundreds of thousands of active users, even though he has been on the platform for less than three months. Collov spokesperson Markk Tong says the company’s AI-designed chatbot has become so popular that it has temporarily shut down the site for fear of overwhelming it.

Much of that attention is driven by viral social media posts and cultural hype around artificial intelligence. Many of these are likely to fade away, and these platforms are powerhouses that will prove to have a long lifespan with consumers. But professional audiences are tuning in as well. The founders of these startups have all seen real estate agents flock to their sites who need to stage properties virtually. Furniture retailers looking for ways to insert products into AI-generated images are also interested.

Interior designers also use these sites in interesting ways. “Dozens of designers contacted me,” he says. “There is a lot of demand among them and they need to decide which potential clients to spend time with. You can ask them if they are on board, start engaging, and get a sense of how serious they are.”

To put it plainly, there will be great gains in this area. Most AI design platforms today offer new users a few free mockups and charge a few bucks for a subscription or credit. This could reach tens of thousands of dollars over time. Looming beyond is the potential from the legions of venture capitalists who have abandoned cryptocurrencies and are now going all-in on AI. investment. One VC gave El Mghari an offer where he was valued at $5 million and he made $1 million. This is a pretty big return on investment considering the fact that he built his RoomGPT in his spare time.

For founders, the promise of AI-powered “killer apps” (tech-industry terminology for extremely powerful and ubiquitous tools) has nearly limitless benefits. “The biggest potential impact I feel is empowering ordinary people who think they are great designers but who are not. It’s a waste of money,’ he says Majumdar. “If you can provide that person with a headless tool that doesn’t cost him $2,000, and maybe he can do it for $200, that’s a huge monetizable market.”

feature
So what can these programs do? Each has its own twist, but the basics are simple. The user uploads an image of the room, fills in a few prompts (such as ‘modern living room’, ‘minimalist kitchen’, etc.) and the algorithm generates a design rendering. An early version of that concept launched last October with Interior AI, but the technology is advancing rapidly.

For one thing, AI is already much better at contracts. The debut version of Interior AI tends to rearrange the room’s architecture (windows, doors, ceiling height, etc.) to achieve the desired look, leaving users with a stylistically correct but non-functional image. It’s gone.

“[The earliest apps] It wasn’t very good,” says El-Mughali. “They gave terrible results. The room that was generated didn’t look like the original room.” We can now play a kind of “tug-of-war” with how one of our engines, Stable Diffusion, works. Now, his AI design tools like El Mghari’s RoomGPT are good at sticking to the original parameters of the space.

Rather than reinventing an entire room, Majumdar’s Reimagine Home takes that control one step further by automatically detecting furniture in a space and replacing specific pieces. “It takes precise control to master,” Majumdar says, noting that most users want to keep at least some of their existing decorations.

CollovGPT is perhaps the most ambitious AI-powered design tool on the market today. Like anything else, it can detect the basic contours of a room and redesign it in different styles. However, its developers have rolled out a variety of bells and whistles. One is product detection. This is a tool that attempts to identify items in AI-generated rooms and generate links to their real-world lookalikes.

Another example is a chatbot intended to tie ChatGPT into the design process.in the demo of BOH, the bot was able to generate a convincing “modern living room” and respond relatively accurately to basic requests such as “remove the sofa” and “make the walls red”. was far from it. CollovGPT couldn’t be more responsive and was prone to silly mistakes (for example, I painted the walls of the room red, as well as the art and the sofa). However, the feature was an interesting glimpse into the future where users can workshop design challenges with his AI in real time.

bug
AI-powered design apps bring state-of-the-art technology to life. This feels like an incredible technological leap. It’s also full of bugs and limitations.

Some of these program weaknesses have nothing to do with artificial intelligence. No site is more than a few months old, and sites often have issues on a daily basis. Pages won’t load, images won’t download, interfaces are clunky, etc. This is very beta software, built on the fly, and appears frequently.

The AI ​​components of these sites also have many limitations. In general, the more ambitious the concept, the greater the margin for error. For example, CollovGPT’s “product identification” tool is a good idea in theory. AI will generate a chandelier, click on it and you will be presented with a purchasable link to a similar one. But in practice, this tool doesn’t really work. I’m having a hard time clearly identifying a specific product. For example, we tend to think of sofas as chairs. Even less, we do not endorse anything that is very similar. (Collov spokesperson Tong acknowledges the tool’s limitations and says that, as with all AI technology, this is a work in progress.

All of these apps also suffer from a lack of differentiation. Each has its own spin, but they all utilize the same AI engine to power their creations, giving them some degree of identity. “We all do the same thing, we use ControlNet,” he says. “There is an element of just waiting. [new AI technology] come out. “

Collov is working to overcome this problem by hosting a “design hackathon”, inviting users to upload design images and tag them with style. Ostensibly a $1,000 contest, the initiative’s purpose is clearly to train the company’s AI. El Mghari is working to develop better tools for designers. Meanwhile, Reimagine Home is working on customizing its models, so we’re experimenting with having the AI ​​write AI-generated descriptions of the rooms.

Over time, developers hope that initiatives like this will help their creations break away from the herd. “AI is the foundational technology and what you build on,” Majumdar said, noting that core technologies such as operating systems and batteries will serve as the foundation for an entire ecosystem of specialized products. It’s likened to a method. “The hard work required to truly understand what the customer needs and to solve it…that is a huge opportunity.”

But despite his enthusiasm for the possibilities, Majumdar is also candid about the challenges. The race to build AI-powered design engines is heating up, but the finish line is still elusive. “It’s a great toy now,” he says. “But there are still chasms to be crossed to be truly useful.

Homepage photo: Generated by Midjourney