3D AI Property
AI-generated 3D properties and interiors, walkable in FPV · with a manual editor and drone view.
Website and API together: enter a property's basic specs and the AI generates its 3D floor plan and interior. You can tweak it by hand, then walk through it in FPV or see it from a drone.
ListenA flat description in plain text. Built an AI 3D-tour generator for property listings. Photos go in, an explorable first-person walkthrough drops out in seconds.
3D AI Property turns a one-paragraph property spec into a generated 3D floor plan + interior, walkable in FPV or viewable from a drone shot. The studio shipped the generation pipeline, the manual tweak editor, and the FPV / drone walkthrough modes.
We honestly didn't think this was possible. Someone types in a property description, and a moment later there's a 3D model of the apartment that you can walk through in first person, like a video game. Our designers can still tweak it by hand, and within two weeks it was hooked into our listing site. Buyers actually call us back because of these tours.
What's on screen
Frame breakdown
- 01User surface
The whole experience the user sees
This frame shows the live product: ai-generated 3d properties and interiors, walkable in fpv · with a manual editor and drone view. Every component is ours · scope, design, code, deploy.
- 02Stack behind the screen
What's powering it: Next.js, TypeScript, OpenRouter
4 stack components run behind this frame · Next.js, TypeScript, OpenRouter drive the visible UI; the rest sit in the data layer. All studio-owned.
- 03What we shipped
AI-generated 3D buildings and interiors
Walkable 3D plan in seconds
- 04Status
Private deploy · under NDA.
Per the client's request the URL stays private · the build, architecture, and lessons can be shared in a scoping call.
How it shipped
Timeline- 01 · BRIEF
What level of fidelity is enough?
Pin the fidelity bar: a buyer needs to feel the layout, not see a CAD-grade render. We scoped to room-level walls, doors, windows, and 10 interior styles · everything beyond was deferred.
- 02 · ARCHITECTURE
Stack decisions before any code.
Decision doc captured the data flow, Next.js, TypeScript, OpenRouter, SQLite role split, and the failure modes we'd handle in v1 vs defer. Cross-service boundaries (where AI ends and the web app begins) were drawn here so neither side leaked into the other later.
- 02 · BUILD
Spec → JSON layout → walkable 3D.
Two-stage AI pipeline: first an LLM converts the prose spec into a deterministic JSON layout (rooms, walls, openings); then a generator renders walkable 3D + interior from that JSON with a chosen style preset.
- 04 · POLISH
Performance, accessibility, and observability.
PSI / a11y / coverage budgets enforced as launch gates. Logging + metrics wired before cut-over · the team can answer 'is it working?' from a dashboard, not a Slack thread. Threat-model checklist signed off before traffic hits the box.
- 03 · SHIP
FPV walk + drone fly + manual edit.
Three viewing modes shipped over the same scene · a manual editor lets buyers and designers nudge walls, swap finishes, switch styles without regenerating from scratch.
What shipped
04- 01Pipeline
Two-stage prose → JSON → 3D
Deterministic intermediate JSON makes the output reproducible and editable · no opaque black-box render.
- 02FPV
First-person walkthrough
WASD + mouse navigation in the browser · the buyer feels the layout instead of guessing from photos.
- 03Drone
Top-down drone fly-through
Cinematic top-down view for marketing assets · embeddable in listing portals.
- 04Editor
Live manual tweak
Move walls, swap finishes, switch interior styles · all without re-running the generator.
From the video
Frame by frame
01FrameBrief input · plain English, scope picker, starter prompts
Brief textarea + scope (Full property / House only / Interior focus) + family / luxury / quirky starter prompts on a 30×30m grid · no CAD knowledge required.
02FrameDesigning · AI lays out 14 rooms across 2 levels
Step 2/2 'Designing' runs · the room layout panel renders the floor plan in 2D coloured cells while the 3D scene grows walls and stairs in real time.
03FrameExterior · style preset 'Realistic' applied
Same property re-rendered with a realistic preset · hip roof, painted exterior, window placement reads correctly. Style swap is a click, not a re-generation.
04FrameStack view · interior breakdown by room
Building stripped to walls + furniture, coloured per room (sleeping / living / wet / utility) · the buyer reads the floor mix from above before stepping into FPV.
THE PROBLEM
- −A flat is hard to picture from photos alone
- −Waiting on an interior designer is slow and expensive
- −Hand-built 3D models take days
WHAT THE CLIENT GOT
- Walkable 3D plan in seconds
- Buyers see the experience before they buy
- Try ten interior styles without a designer
WHAT WE DELIVERED
- +AI-generated 3D buildings and interiors
- +Manual editor for fine-tuning
- +FPV and drone views · actually walkable
- +Switch styles with one click
STACK
- Next.js
- TypeScript
- OpenRouter
- SQLite
RELATED READING
- AI solutions · Websites, web apps & online shops · Cybersecurity · Custom software · everything elseQ3 2026 roundup: what shifted, what we shipped, what brokeThree months in. SZEP 2.0 live, NAV v3 cutover, AI Act enforcement, OWASP LLM Top 10 v2. Hard numbers, one strong opinion on the consulting tier.
- AI solutions · Websites, web apps & online shops · Custom software · everything elseQ2 2026 roundup: what shifted, what we shipped, what brokeFour months in. Eleven shipped projects, real before/after numbers, one strong opinion on what the consulting tier got wrong this quarter.
- Custom software · everything else · AI solutionsn8n vs Make vs custom code: 2026 automation stackNo-code automation is brilliant until it isn't. Here's the line where n8n / Make stop saving money and custom code starts - and how to tell which side you're on.
- AI solutionsAI agent pricing 2026: what an autonomous agent costsAn AI agent is not a chatbot with extra steps - it takes actions, and that changes the bill. Here are the real 2026 ranges and what drives them.