InteriorAI — An AI That Redesigns Your Room From a Photo, Making $40K/Month. The First of the "AI × Existing Industry" Mass-Production Line
InteriorAI, which generates interior design proposals from an uploaded photo of your room, makes $40K per month (about ¥6 million). Like PhotoAI, it is one of the products Pieter Levels mass-produced in the early days of the 2022 AI image-generation boom — a "replace an expensive existing service with AI" play.
This is a translation of the Japanese original. The Japanese version and cited primary sources are authoritative for all figures.
Yen conversions in this article are rough estimates at $1 = ¥150.
The Numbers at a Glance
| Item | Figure |
|---|---|
| Monthly revenue | $40,000 (about ¥6 million) / figure published in 2025 |
| Launched | 2022 (early in the AI image-generation boom) |
| Team | Pieter Levels (zero employees) |
The Business and Its Lessons
InteriorAI is a service that generates interior design proposals in styles like “Scandinavian” or “minimal” from a photo of a room. It replaced a domain where your only options were paying an interior coordinator tens of thousands of yen or wrestling with your own taste — with an AI costing a few dozen dollars.
When you replace a “service that’s expensive when done by humans” with AI, the price gap itself becomes the value proposition. It is overwhelmingly cheap compared to professional coordination fees — the same structure as PhotoAI’s “1/100th the price of a photo studio”. It shows that the shortcut for planning AI products is to start not from the technology but from a “list of expensive existing human services.”
It is also one piece of a strategy of mass-producing multiple products from the same technical foundation. From Stable Diffusion, the same underlying technology, Levels ran PhotoAI (people) and InteriorAI (spaces) in parallel. By reusing the cost of learning the technology across multiple bets, he built a structure where even a “4 hits out of 70” batting average has positive expected value.
How to Plan an “AI × Expensive Existing Service” Product
InteriorAI’s conception did not start from a tech demo. He chose a job — “interior coordination” — that satisfies three conditions: (1) clear existing demand, (2) expensive when you hire a human, and (3) the deliverable is fully contained in an image — and then applied Stable Diffusion to it. These three conditions can be used as-is as selection criteria for products individuals build with image-generation AI. Jobs whose deliverables are not fully contained in an image (e.g., legal advice) carry heavy accuracy liability and are not suited to individuals.
The other design decision is that the same feature works for both B2C (redecorating your home) and B2B (virtual staging of property photos for real estate agents). The latter in particular replaces an existing paid task — “placing furniture into photos of vacant rooms” (tens of thousands of yen per job) — and business usage rarely churns. The same AI feature is structured so the wallet is tested in B2C and fattened in B2B.
Its Place in the Mass-Production Portfolio
Within Levels’s portfolio (PhotoAI at $132K/month, RemoteOK, Nomad List, and more), InteriorAI’s $40K/month puts it in the “second tier.” What matters is that this is maintained without a dedicated operator. Because the technical foundation, payments, and infrastructure are shared across products, the cost of keeping a second-tier product running is close to zero. Running multiple indie products only becomes a portfolio when each one is designed to run on part-time attention — with a setup where each product demands your full effort, a mass-production strategy cannot work.
Related Cases
Sources
This article summarizes and analyzes the public sources above. Please refer to the primary sources for details.