RoomCart

AI virtual staging built for real estate agents. Upload one photo of an empty room, and in a couple of minutes you get a fully staged listing image, video, and a shoppable page that captures leads

Comment

Inspiration

Every agent knows staged homes sell faster and for more — but professional staging runs $1,500–3,000 per listing and takes days. For the individual agents and small brokerages who make up most of the market, that's simply out of reach, so their listings go up empty and sit longer. We kept coming back to one idea: the biggest financial decision in most people's lives shouldn't hinge on whether an agent can afford a furniture rental. If AI could produce professional-looking staging for a few dollars, in minutes, we could hand small-business agents the same edge the big firms have.

What it does

RoomCart is AI virtual staging for real estate agents. An agent uploads one photo of an empty room and gets back, in minutes:

  • a fully staged, listing-ready image,
  • a short cinematic "reveal" video, and
  • a branded showcase page where every piece of furniture is shoppable and every buyer inquiry is captured as a lead.

No stager, no rented furniture, no scheduling — and no human in the loop. It's live on the web and as native iOS and Android apps.

How we built it

The whole product is an AI pipeline on Google Cloud:

  • Gemini (Vertex AI) reads the room's geometry, lighting, and style and generates the photorealistic staged image — the core creative decision.
  • A matching step maps the generated furniture to real, purchasable products, making the image shoppable.
  • Veo (Vertex AI) turns the staged room into a cinematic reveal video.
  • Cloud Run hosts the app and orchestrates the pipeline; Firebase (Auth + Firestore) handles accounts, an append-only credit ledger, listings, and leads; Cloud Storage holds the media.

It's a pnpm monorepo: a Next.js web app, an Expo (React Native) mobile app, and a shared TypeScript core for design tokens and domain types. On top we built shoppable hotspots, per-listing analytics, agency/team plans with a shared credit pool, and a billing system spanning Stripe (web) and Apple In-App Purchase via RevenueCat (iOS).

Challenges we ran into

  • Making the AI reliable in production. Getting Gemini to stage a room believably, matching that furniture to real products, and generating a coherent Veo reveal — as one continuous pipeline — took real iteration.
  • The App Store gauntlet. We went several rounds: account deletion had to be genuinely in-app (Guideline 5.1.1), the camera-permission wording had to change, and Guideline 3.1.1 forced us to build native In-App Purchase — including the chicken-and-egg where consumable products won't load in the sandbox until they've been submitted with a build.
  • The economics of AI media. A reveal video costs real compute, so we designed a credit system and an append-only ledger to keep unit costs honest and margins predictable.
  • One product, three billing regimes. Web (Stripe), iOS (Apple IAP), and Play — each with different rules — over a single shared credit balance.

Accomplishments that we're proud of

  • A complete, production AI pipeline — staging and video — running live on Google Cloud, where AI does 100% of the production work.
  • A real full stack shipped: web plus native iOS and Android, not a prototype.
  • The details that make it a business, not a demo: shoppable carts with affiliate income, branded lead-capturing showcases, team plans, analytics, and compliant billing across both app stores.

What we learned

  • How to build a business where AI is the workforce — designing every step so the marginal cost of one more staged listing is essentially compute, and it scales without adding staff.
  • App-store compliance is a product surface, not an afterthought — account deletion, purchase flows, and store metadata all shape the build.
  • AI-generated media has a real COGS, and the product design has to respect it.

What's next for RoomCart

  • Go-to-market: rolling out to multi-agent agencies and top-producing teams in the US and Malaysia with founding-agent offers.
  • More AI: additional design styles, richer reveal motions, and deeper room understanding.
  • Google Play Billing to complete native purchases on Android, plus broader international catalogs.
  • A creator + reseller layer: enabling staging-as-a-service resellers to deliver AI staging to their local markets — turning RoomCart into economic opportunity beyond the founding team.

Built With

Share this project:

Updates