Do not start with only a yes-or-no question
An address alone may lead to confusion with a similar place, neighbouring zone or outdated material. Establish the parcel, boundaries of the enquiry, document dates and intended use first. Do not present a precise conclusion from ambiguous inputs.
Separate confirmed official facts, map-based reference information, site checks and questions for authorities or specialists. The useful result is a next-action list, not an instant pass/fail score.
Pair desk research with its source
Japan’s MLIT Real Estate Information Library lists price, hazard and planning information among its content. Check the source, date and geographic scope of each layer. A map display is an initial research aid, not final evidence of a parcel boundary or legal determination.
| Task | AI assistance | Human verification |
|---|---|---|
| Documents | Compare fields, dates and figures | Originals, parcel scope and transcription |
| Land use and hazards | Organise public sources and open questions | Current local rules, applicability and evacuation |
| Prices | Build a like-for-like comparison | Asking versus transaction price, date, area and use |
| Site visit | Prepare questions and checklists | Levels, roads, services, noise and odours |
| Options | Identify differing assumptions and missing evidence | Design, construction, permissions, quotes and decisions |
Keep “unverified” visible in the report
The following is an illustrative record for a hypothetical site, not a completed KODACHI assessment.
| Item | Example record | Next step |
|---|---|---|
| Zoning | Public map consulted; parcel location not yet reconciled | Confirm parcel and use with the municipality |
| Road | Width stated in sales particulars; legal status unknown | Obtain road and rights records |
| Drainage | No evidence available | Check connection conditions with the relevant office/provider |
| View | Sales photo; camera height unknown | Photograph proposed seating positions on site |
| Earthworks | No estimate | Obtain specialist review based on surveys |
If AI fills an empty field with a guess, the report looks fuller without becoming more reliable. Recording that evidence is unavailable helps prevent an assumption becoming a decision.
Turn questions into inspections
A connected line on a map does not establish vehicle turning room, unloading space or night-time wayfinding. Likewise, a view depends on camera height and neighbouring conditions. Convert questions into specific tasks: photography, surveys, service checks and authority enquiries.
Record photo direction and position. Avoid sending unnecessary faces, registration plates, contracts or guest information to an external AI service. Check its data terms and the organisation’s approval scope, using only the data needed.
Review the evidence before using the conclusion
Check that links exist and support the statement, that nearby sites have not been confused, and that update/verification dates are recorded. Keep asking prices separate from completed sales. Unknown costs must not become zero, and predicted occupancy must not become historic performance.
Retain access to original records, rather than keeping only the AI summary. Record the assumptions that would change the decision so the report can be updated when new evidence arrives.
Prepare for an initial consultation
Bring the location/parcel, available drawings or sales particulars, intended use, budget scope and timing. If there is no chosen plot yet, begin with areas and priorities. KODACHI’s free land-use and AI-research consultation accepts enquiries from Okinawa, Tokyo and elsewhere in Japan. Confirm the consultation scope first; surveys, design and specialist investigations may require separate arrangements and fees. Architectural design is undertaken by the introduced practice.
Sources and further reading
Published / sources checked: 9 November 2026