Cabocia Inc. (Headquarters: Katsushika-ku, Tokyo; Representatives: Riku Koike, Tomohito Nakato) will officially release its real estate database "FUDOSAN DB" on July 6, 2026. Following a trial period, the service will be launched as a platform for searching and analyzing data from the Ministry of Land, Infrastructure, Transport and Tourism's real estate transaction price information, land price surveys, and J-REIT (Japan Real Estate Investment Trust) disclosures. API requests currently exceed 1 million per month.
The supported data includes three main pillars: 6.26 million real estate transaction price records (2005-2025), 44 years of land price survey time-series data (1983-2026), and over 100,000 J-REIT property data points (67 investment corporations, 2002-2026). Area reports for 1,920 municipalities nationwide can be viewed without membership registration. The rent estimation AI recorded a MAPE (Mean Absolute Percentage Error) of 1.6% in urban areas and 2.0% nationwide in verification data, and can be directly called from AI assistants like Claude, ChatGPT, and Copilot via 39 REST API endpoints and 33 MCP tools.
To commemorate the official release, a special offer will be available for those who register for an account (free) by July 20, 2026: a permanent ¥1,000 monthly discount on the Pro plan (from ¥2,490/month to ¥1,490/month). The discount will be automatically applied upon upgrade.
URL: https://fudosandb.jp/
Official Release of FUDOSAN DB
Publicly available real estate data is dispersed and difficult to compare.
The Ministry of Land, Infrastructure, Transport and Tourism's Real Estate Information Library (reinfolib) provides 35 types of APIs for public data on real estate nationwide, including transaction prices, land price surveys, and urban planning data. Additionally, J-REIT companies periodically disclose property acquisition prices, appraisal values, rents, and occupancy rates in their financial reports.
While this information is available for free, it originates from different sources and is in various formats. To research transaction prices, one would use the Real Estate Information Library; to track land price trends, the land price surveys; and to check J-REIT property details, each company's disclosures. When wanting to cross-reference multiple data sources, users must individually collect the data, standardize its format, and compare it.
FUDOSAN DB integrates these public data sources into a single database, offering access through both web searches and direct calls from AI (such as Claude and ChatGPT).
Provided Data and Features
Key Data Assets: 6.26 Million Transaction Prices, 44 Years of Land Prices, Over 100,000 J-REIT Properties
Data Assets
Data
Scale
Real Estate Transaction Price Information
6.26 million records (Ministry of Land, Infrastructure, Transport and Tourism, 2005-2025)
Land Price Survey
44 years of time-series data (1983-2026)
J-REIT Property Data
Over 100,000 records (67 investment corporations, 2002-2026)
Supported Areas
1,920 municipalities nationwide
The system integrates the 35 types of APIs from the Ministry of Land, Infrastructure, Transport and Tourism's Real Estate Information Library, allowing access to all data with a single API key and a single MCP connection.
Rent Estimation AI
Estimates monthly rent anywhere in Japan based on address and area, among other conditions. The estimation error (MAPE, Mean Absolute Percentage Error) in verification data is 1.6% in urban areas and 2.0% nationwide. While training combines public data with actual rent data, the input data referenced for inference (actual estimation calculations) is limited to public data, ensuring a system that does not refer to external non-public data on a case-by-case basis.
Rent Estimation AI Accuracy: 1.6% MAPE in Urban Areas, 2.0% Nationwide
Yield Simulation
Input property information to simulate gross yield, net yield, and cash flow based on estimated rent.
Web Dashboard (No Registration Required, Completely Free)
- Nationwide Color Map (Visualization of land prices and population dynamics) - Area Details (Land price trends, population, disaster risk) - Rankings (Multiple categories including land prices and transaction volume) - Cross-J-REIT Search & Appraisal Comparison
Designed for Direct AI Use
Supports both REST API (39 endpoints) and MCP (33 tools), allowing data to be called conversationally from AI assistants like Claude, ChatGPT, and Copilot. Connection procedures are published on the developer page.
3 Steps to Connect to Claude, ChatGPT, Copilot
From Individual Investors to Corporations — FUDOSAN DB as Data Infrastructure
FUDOSAN DB is designed as a database to support research and analysis based on public data, without selling properties itself. Property information will not be presented for sales purposes.
Area reports for 1,920 municipalities nationwide can be viewed without membership registration. By registering for the free Free plan, users can access features such as rent estimation, yield simulation, and cross-J-REIT searches within usage limits. It can be used by individual real estate investors as a comparative tool during property consideration.
For corporations, customized plans are available based on data volume and usage. These are intended for integration into data analysis platforms and for cross-research purposes in asset management (AM) and private funds.
Furthermore, the information provided by FUDOSAN DB is for research and analysis purposes based on public data and does not constitute investment advice.
Differences from Existing Research Methods
Previously, methods for researching public real estate data were broadly divided into two categories:
1. Directly referencing individual pages or APIs for each data source, such as the Ministry of Land, Infrastructure, Transport and Tourism's Real Estate Information Library or various portal sites. While comprehensive, the different formats across sources require pre-processing by the user for cross-comparisons. 2. Business tools for real estate brokerage and appraisal. These services primarily focus on calculating estimated sale and purchase prices by inputting a specific address and are used in real estate sales.
FUDOSAN DB occupies a different position from both. It does not offer individual property appraisal or brokerage functions but is designed as a database that structurally integrates public data for cross-referencing, providing access through both web searches and AI connections.
Comparison Axis
Referencing Information Sources Individually
Real Estate Business Tools
FUDOSAN DB
Data Scope
Individual to each source
Primarily individual property appraisal
Cross-referencing transaction prices, land prices, and REITs
Cross-Comparison
Requires user pre-processing
Not applicable
Completed within the database
AI Utilization
Not supported
Not supported
REST API & MCP Support
Source Attribution
Depends on each source
Mostly non-public
Explicitly states primary sources like MLIT and EDINET
Official Release Special Offer
To commemorate the official release, a special offer will be available for those who register for an account (free) by July 20, 2026: a permanent ¥1,000 monthly discount on the Pro plan (from ¥2,490/month to ¥1,490/month). The discount will be automatically applied upon upgrade.
Plans
Plan
Monthly Fee (Tax Incl.)
Request Limit (Requests/Day)
Free
¥0
100/day
Pro
¥2,490 (¥1,490 for registrations by 7/20)
500/day
Enterprise
Inquire
Custom
*The discount will be applied monthly for the duration of your Pro plan subscription (until cancellation). Details on eligibility conditions can be found on the registration page.
Why Now?
As the use of AI assistants for information retrieval expands, there is a growing need to provide real estate data not only as human-readable web pages but also as structured data that AI can directly reference. FUDOSAN DB addresses this shift by being designed for direct calls from AI assistants via both REST API and MCP.
The rent estimation AI is also designed with input data for inference limited to public data, intending to facilitate source explanation even when provided to third parties via AI assistants.
Cross-J-REIT Analysis via Conversation with AI Chat (Claude, ChatGPT, Copilot, etc.) using MCP
Future Developments
Expansion of data related to zoning and urban planning is planned.
Service Overview
Item
Details
Service Name
FUDOSAN DB
URL
https://fudosandb.jp
Data Sources
Ministry of Land, Infrastructure, Transport and Tourism Real Estate Information Library (reinfolib), J-REIT disclosure data, etc.
Key Data
6.26 million real estate transaction price records, 44 years of land price survey data, over 100,000 J-REIT property data points
API
REST API (39 endpoints) + MCP (33 tools)
Pricing
Free ¥0 / Pro ¥2,490 / Business ¥29,800 / Enterprise Custom
Company Profile
Item
Details
Company Name
Cabocia Inc.
Location
5-14-8 Kanemachi, Katsushika-ku, Tokyo
Representative
Representative Directors Riku Koike, Tomohito Nakato
Business Activities
Listed Company Database "EDINET DB"
Real Estate Information Database "FUDOSAN DB"
Policy Information Database "SEISAKU DB"
Disaster Prevention Information Database "BOSAI DB"
Development and operation of Japan's AI-first data infrastructure
Inquiries about FUDOSAN DB
FUDOSAN DB
https://fudosandb.jp/
FACT BOX
- Source: PR TIMES
- Category: 製品リリース
- Organizations: ChatGPT / Claude / Copilot