Structuring over 6.2 million MLIT real estate data points. Free release of API 'FUDOSAN DB' with AI rent estimation
Kaboshia Inc. launched the beta version of 'FUDOSAN DB' on April 1, 2026. It structures massive public real estate data and provides an AI rent estimation engine, accessible directly via ChatGPT.
📋 Article Processing Timeline
- 📰 Published: April 6, 2026 at 17:30
- 🔍 Collected: April 6, 2026 at 09:00
- 🤖 AI Analyzed: April 21, 2026 at 03:17 (354h 17m after Collected)
Kaboshia Inc. (Headquarters: Minato-ku, Tokyo, Representative: Riku Koike) released the beta version of 'FUDOSAN DB' on April 1, 2026. This platform structures data primarily from the Ministry of Land, Infrastructure, Transport and Tourism's real estate information library (reinfolib) and provides it via a REST API and AI connector.
In addition to over 6.2 million real estate transaction prices and 130,000 land price surveys, it is equipped with a machine learning-based rent estimation engine (1.8% error rate), rent fluctuation prediction, and revenue simulation. It can connect directly from ChatGPT or Claude via a connector, allowing users to analyze real estate data using natural language.
Web browsing requires no registration and is completely free. API keys are issued immediately without an application process, and there is a free tier. During the beta period, Pro-equivalent features (1,000 requests/day) are available for free, and special benefits are prepared for the official release for those who register during the beta.
https://fudosandb.jp
■ Real estate data is published for free. But no one is able to master it.
The MLIT's real estate information library (reinfolib) freely publishes 35 types of data via API, including nationwide transaction prices, land prices, and urban planning data. Over 4,000 people have applied for API keys, generating 4 million requests per month.
However, when actually trying to integrate this data into a product, there are two walls, especially for individuals.
Wall 1: Preprocessing is too heavy
The tile API returns PBF (Protocol Buffer Binary Format) instead of JSON, requiring decoding and coordinate conversion. The area of the same transaction differs by 5 to 15 square meters between the CSV and API, meaning they cannot be integrated with a simple JOIN. Preprocessing alone takes weeks to months. During that time, actual product development stops. Furthermore, issuing an API key takes several days or more.
Wall 2: Structured data is expensive to buy
It is common to think, 'If I can't preprocess it myself, I should just buy processed data.' However, domestic real estate data API services are mainly for corporations, starting at 50,000 to 100,000 yen per month. Everywhere starts with 'Please contact us first,' and there are no self-serve services ready for immediate use. APIs and structured data at a price point that individual developers, researchers, and startups can easily test have not existed until now.
Two years have passed since the reinfolib API was released. Even though more than 4,000 people have obtained keys, a service that overcomes these two walls to integrate all 35 types of data and provide it as structured data has not been born. The reality is that valuable public data covering everything from nationwide transaction prices to land prices and urban planning is not fully utilized due to the heavy burden of preprocessing and the wall of cost.
FUDOSAN DB solves both of these.
35 types of data
In addition to over 6.2 million real estate transaction prices and 130,000 land price surveys, it is equipped with a machine learning-based rent estimation engine (1.8% error rate), rent fluctuation prediction, and revenue simulation. It can connect directly from ChatGPT or Claude via a connector, allowing users to analyze real estate data using natural language.
Web browsing requires no registration and is completely free. API keys are issued immediately without an application process, and there is a free tier. During the beta period, Pro-equivalent features (1,000 requests/day) are available for free, and special benefits are prepared for the official release for those who register during the beta.
https://fudosandb.jp
■ Real estate data is published for free. But no one is able to master it.
The MLIT's real estate information library (reinfolib) freely publishes 35 types of data via API, including nationwide transaction prices, land prices, and urban planning data. Over 4,000 people have applied for API keys, generating 4 million requests per month.
However, when actually trying to integrate this data into a product, there are two walls, especially for individuals.
Wall 1: Preprocessing is too heavy
The tile API returns PBF (Protocol Buffer Binary Format) instead of JSON, requiring decoding and coordinate conversion. The area of the same transaction differs by 5 to 15 square meters between the CSV and API, meaning they cannot be integrated with a simple JOIN. Preprocessing alone takes weeks to months. During that time, actual product development stops. Furthermore, issuing an API key takes several days or more.
Wall 2: Structured data is expensive to buy
It is common to think, 'If I can't preprocess it myself, I should just buy processed data.' However, domestic real estate data API services are mainly for corporations, starting at 50,000 to 100,000 yen per month. Everywhere starts with 'Please contact us first,' and there are no self-serve services ready for immediate use. APIs and structured data at a price point that individual developers, researchers, and startups can easily test have not existed until now.
Two years have passed since the reinfolib API was released. Even though more than 4,000 people have obtained keys, a service that overcomes these two walls to integrate all 35 types of data and provide it as structured data has not been born. The reality is that valuable public data covering everything from nationwide transaction prices to land prices and urban planning is not fully utilized due to the heavy burden of preprocessing and the wall of cost.
FUDOSAN DB solves both of these.
35 types of data
FAQ
What is FUDOSAN DB?
It is an API platform that structures over 6.2 million real estate data entries from the Ministry of Land, Infrastructure, Transport and Tourism, and provides AI rent estimation directly usable from ChatGPT and others.
How much does it cost to use?
Web browsing is completely free and does not require registration. The API has a free tier, and during the beta period, Pro-level features are available for free.
What problems does it solve?
It addresses the heavy data preprocessing required when handling public data and the high monthly cost of 50,000 to 100,000 yen for purchasing structured data.