Furnishing an apartment usually means a browser full of open tabs, a tape measure, and some uncertainty about whether the sofa will fit the space it was meant for. The buying is spread across dozens of retailer sites, and the shopper often doesn’t know whether the pieces actually work together until they’re delivered.
A free iOS app called elephantroom was built to reduce that uncertainty. Co-founded by architect Lora Fahmy and AI engineer Alessandro Duico, who serves as chief technology officer, the app creates a full room layout in seconds using products from store catalogs, each sized to the centimeter. Shoppers can set their preferred style, budget, and brand by choosing from options generated by the assistant. They can then send the entire plan to a shopping cart.
Where AI Design Apps Fall Short #
The global furniture market is worth roughly $786 billion, yet much of the buying remains stubbornly analog and disjointed, closer to the pre-internet way of shopping than to booking a flight or ordering groceries.
The result is a familiar stall. Shoppers hit a version of designer’s block: they simply can’t picture the finished room, and can’t be sure if a given piece works well in their space. A wave of AI design apps promised a fix, generating photorealistic rooms from a single prompt. In practice, those renders can have limitations. Furniture may be generated rather than based on purchasable products, and even when a real model is used, its dimensions may not be represented precisely enough to show how it could fit within a room.
Duico believes this limitation is structural, built into the technology. “The problem is that it’s never going to be physically accurate, at least for the foreseeable future, because it doesn’t know the space the picture represents or the measurements of the furniture,” he says. Most apps, he explains, lean on technology that paints furniture into a photo, reverse-engineering that final result to guess where each piece sits.
The result looks convincing and means little to someone trying to order a couch.
The Founders Teaming Up to Solve This Problem #
Lora Fahmy, elephantroom’s co-founder, first came to the problem as a practitioner. A computational architect, Fahmy ran into the same wall while furnishing a new home in Milan, finding the process fragmented and slow even with design software at her disposal. If the tools built for the trade could not close the gap, the ones aimed at ordinary shoppers stood no chance.
Alessandro Duico arrived from a different angle. He was returning to Italy after sharpening his technical outlook in the San Francisco startup scene, having studied computer science at Politecnico di Milano and completed a master’s at TU Delft in the Netherlands. He also brought hands-on product experience, which taught him how to take AI to a mass audience.
An Italian venture builder, Liquid Factory, spotted the fit and brought the two together. The pairing was complementary by design, matching Fahmy’s command of architectural space with Duico’s work on incorporating AI into a system able to reason about geometrical rules.
Inside elephantroom #
Now live on the App Store, elephantroom seeks to make the design experience frictionless and intuitive. After first signing up, a user sets a room, then settles on style, layout, and budget with visual accompaniments for all options created by AI. Instead of painting furniture into a photo, the app builds the room in a real 3D engine with the objects scaled to true real-world size, so what appears on screen corresponds to something that can actually be placed and bought.
That foundation is what makes the speed possible. The app can create hundreds of workable arrangements in about two seconds, and according to the company, the furniture shown on screen is drawn from products available through retailers, alongside smaller local makers, not the generic placeholder pieces that populate most design tools.
From there, the plan stays editable. Users can ask to move things around, change styles, swap brands, or reset the budget, then send everything to a cart with direct links to buy. A separate photo mode layers realistic rendering on top of a detailed 3D layout designed to provide both a visual representation and a practical room plan. Duico sums up the difference plainly: “You’ll get real 3D designs matched to real products you can instantly buy. No dummy furniture, no complicated tools, no hidden paywalls.”
The Technology Behind the Product #
According to elephantroom, the platform’s approach to accuracy relies not only on image generation but also on how the team digitizes retail catalogs at scale. Product pages from major brands are converted into 3D objects complete with dimensions, prices, and materials, a job measured in hours rather than the days manual modeling would take. Automated scripts then resize each model to the measurements the retailer actually declares, to help the digital representation reflect the physical product more closely.
That resizing runs on two separate steps. An image-to-3D model turns each product photo into a rough model, right in form but not in size, while a language model reads the catalog text for the piece’s declared depth and width. “Our LLM goes through every catalog, works out a product’s depth and width, and converts it to different units,” Duico explains. “That’s how we have a catalog of 3D models with true-to-scale measurements.”
That library feeds a layout algorithm treating each product’s real dimensions as hard limits, arranging pieces around the walls and the walking space a room needs to function. A wardrobe can’t quietly grow into a wall to make the picture look better, reducing the likelihood of shoppers encountering unexpected sizing differences upon delivery.
Alessandro Duico and Lora Fahmy built elephantroom to answer a plain question that expensive design software and prompt-based apps both leave open: will the room actually work when the furniture arrives? By pairing an architect’s read on space with real catalog products sized to the millimeter inside a 3D engine, the app turns furnishing a home from guesswork into an AI-enhanced plan that shoppers can review before ordering products from the screen.