-
-
Room awareness: room name, size and features inside VR
-
Walking through the pre-construction concept home at full scale
-
DWELSAV AI answers about the current room, with sources
-
Living room of the concept home in WebXR on Meta Quest
-
Compare 3 homes with personalized fit scores
-
Decision Board: fit scores, trade-offs and next steps on the web
Inspiration
At the end of 2021, one question stayed with me: could someone experience a home in another country without traveling there?
I was learning about real estate, virtual reality, and the metaverse. In 2022, I wrote the idea down. That question eventually became DWELSAV.
Buying a home from far away is difficult. Listings provide photos, prices, and basic facts, but they rarely communicate how a space feels or how well it fits a buyer's life. People relocating across cities or countries often make one of their biggest decisions with fragmented information.
DWELSAV was created to bring spatial exploration, property intelligence, and the decision process together.
What it does
DWELSAV is a property decision platform for Meta Quest and the web.
- Spatial home exploration: Buyers can begin at the front door and explore a home at real scale in WebXR using Meta Quest hand tracking.
- Room awareness: The experience identifies the room being visited and presents relevant information about the space.
- Ask DWELSAV inside VR: Buyers can ask questions about the current room or the property without leaving the immersive experience. Answers show their source, and DWELSAV says when information is missing. On the web, DWELSAV AI also works in English, Spanish, and French.
- Property intelligence: DWELSAV brings together listing information, location data, buyer preferences, and personalized analysis.
- Three-home comparison: The Decision Engine compares three properties using personalized fit scores and explains trade-offs, missing information, and next steps.
- Pre-construction exploration: The demonstration home shows how buyers can experience a property before it is physically built.
DWELSAV provides information and explanations without making the decision for the buyer.
How we built it
The spatial experience was built with WebXR and Three.js and deployed through Vercel. The demonstration home was modeled in Blender and exported as an optimized glTF asset with KTX2 textures.
Room volumes allow the experience to understand where the buyer is inside the home. DWELSAV AI connects property information, buyer preferences, location intelligence, and the Decision Engine through one unified assistant. Supabase supports authentication, buyer profiles, listings, and application data.
The prototype was tested in the Meta Quest Browser with hands-first interactions, including pointing, pinching, teleportation, interface controls, and an in-VR keyboard.
Challenges we ran into
One major challenge was optimizing a detailed 3D home for a comfortable browser-based experience on Meta Quest.
Another was designing interactions that work naturally with hands. Navigation, turning, interface selection, and typing all needed to remain understandable without requiring controllers.
We also had to make the AI useful without allowing it to invent property facts. DWELSAV distinguishes between available information and missing information and keeps the final decision with the buyer.
The final challenge was combining immersive exploration and property analysis without overwhelming the user. The interface keeps the home visible while making DWELSAV available when needed.
Accomplishments that we're proud of
- Built a complete hands-first spatial home tour for Meta Quest Browser
- Created a room-aware experience that understands which space the buyer is visiting
- Connected spatial exploration with property questions and personalized intelligence
- Built a Decision Engine that compares three homes and explains their trade-offs
- Created an optimized pre-construction concept home that buyers can explore before it is built
- Brought the experience together in a focused demonstration that can be completed in only a few minutes
What we learned
We learned that an immersive real estate experience needs more than a realistic 3D home. Buyers also need context, clear information, and help understanding how a property fits their lives.
We learned how important performance, comfortable navigation, and focused interface placement are on Meta Quest. Hands-first design requires every action to be intentional and easy to understand.
We also learned that AI should support a real estate decision, not replace the buyer or the real estate professional.
What's next
- A shared Spatial Agent Room where a buyer and real estate professional can explore the same property remotely
- Direct buyer-agent messaging and tour requests
- Voice questions inside VR and additional language support
- Smarter property alerts based on each buyer's profile
- Expanded tools for pre-construction properties and real estate developers
- Controlled market-by-market pilots with verified real estate professionals
The target pilot launch is early 2027.
The buyer decides. The agent advises. DWELSAV connects the journey.## Inspiration
Built With
- ai
- blender
- claude
- gltf
- hand
- javascript
- ktx2
- meta
- quest
- spatial
- supabase
- three.js
- tracking
- typescript
- vercel
- webxr