The Engawa Journal
Practical field notes for international buyers: how to search, what to verify, when to ask for help, and how to spend time in a place before making a long-term decision.
Featured field note
107 published guides
Japan's revised condominium laws took effect on April 1, 2026. For buyers, the practical issue is not memorizing the statute—it is checking whether an older building's governance, reserves and regeneration plans have caught up.
Japan broadened the FEFTA real-estate reporting rules on April 1, 2026. For non-resident buyers, even a zero-yen acquisition can be reportable within 20 days.
Engawa MCP is a tool layer for Japanese property research. An AI assistant can search Engawa's catalogue, open a full listing record, compare up to five properties, estimate closing costs, list prefectures with inventory and surface the concierge handoff without pretending to be the broker.
Japan is unusually well suited to an MCP layer because the hard part is not only finding a house. It is reconciling fragmented listings, Japanese terminology, location context, old-building risk and transaction costs while knowing which facts still require a human check.
AI-native search means the first interface can be a sentence such as 'find me a renovation project near rail with a large plot,' while the underlying system still uses deterministic filters and source-linked listing records.
A property portal and an MCP server solve different interface problems. The portal is better for visual exploration and serendipity. MCP is better when the buyer has a multi-constraint brief and wants the assistant to iterate across search, detail, comparison and cost calculations.
A REST API can power an app perfectly well. MCP matters when the caller is an AI model because the tool names, descriptions and input schemas become part of how the model decides what to do next.
The useful pattern is a funnel. Start broad with structured filters, then spend context and attention only on the listings that survive. The assistant should not open fifty full records when twelve compact rows are enough to choose the next step.
Data becomes agent-ready when a model can tell what a field means, where it came from, whether it is missing and what action should follow. A large JSON dump is not enough.
In property research, 'unknown' is often the correct answer. The assessed value may not be in the listing. A hazard lookup may not resolve. The property may have no survey. An AI system should carry those absences forward as data.
Cross-border buyers face more than translation. They need to reconcile currency budgets, unfamiliar building standards, prefecture names, local transport, taxes and the practical fact that buying a house does not solve residency or remote-management questions.
Engawa should not claim to be the first real-estate MCP in Japan. By 2026, MLIT had published a geospatial MCP, FUDOSAN DB had an MCP workflow, Japan Property Research exposed a property-research MCP, and other projects covered real-estate intelligence.
MCP does not create investment returns by itself. It creates leverage in the research process: more hypotheses can be tested, weak deals can be rejected earlier and acquisition costs can be modeled before an investor becomes attached to a listing.
A ¥1M house is not automatically undervalued. It may be appropriately priced for condition, access, title complexity or location. Engawa MCP is useful for finding candidates whose known attributes fit a thesis, then separating the thesis from the facts still needed to prove it.
A flip screen should answer one question early: is this property worth paying to investigate? MCP is good at the cheap first pass because it can apply the same filters and cost logic across many listings.
Yield screens fail when the inputs look more precise than they are. Engawa can provide an area rent benchmark where available and can estimate closing costs, which is useful for comparison, but the investor must still validate achievable rent, vacancy, management and renovation.
Land is attractive to AI screening because the structured fields look simple: price, area and location. The danger is assuming those fields capture usability. Road frontage, zoning, services, topography and local planning can determine whether cheap land is useful at all.
A second-home decision is not purely financial. The property has to be reachable, manageable from abroad and pleasant enough to justify repeated use. An AI assistant can keep those constraints together while screening listings.
A percentage rule of thumb works worst at the bottom of the market. Fixed professional fees, registration and taxes can make a ¥1M or ¥3M purchase much more expensive relative to the headline price than a normal urban transaction.
Comparison is where an AI assistant can be more useful than a portal. The model can keep the buyer's priorities stable and explain why one listing wins on land, another on building age and another on price.
A watchlist is useful when it records why a property is interesting and what would disqualify it. MCP makes that easier because the assistant can rerun the same filter logic and store a consistent summary in the conversation or the user's own system.
AI is most useful in diligence as an organizer. It can gather the catalogue facts, compare them consistently, identify explicit gaps and produce a property-specific request list for the agent, inspector, scrivener or tax adviser.
For agents, the most useful MCP promise is not automation of the transaction. It is a better front door. The buyer can arrive with a budget, target area, shortlist and list of unresolved questions instead of asking the agent to perform free-form discovery from zero.
A better lead is not simply a buyer who came through AI. It is a buyer whose intent is legible. The agent should know what they want, what they can spend, which properties they considered and what problem still requires the agent.
An AI product becomes agent-hostile when it overclaims what it can verify, captures the entire client relationship and treats the agent as a commodity at the last step. Agent-friendly design does the opposite: it is explicit about limits and creates a cleaner handoff.
Prequalification here does not mean mortgage approval. It means determining whether the buyer's brief can produce plausible inventory and whether the buyer understands the practical constraints of a Japanese purchase.
The expensive enquiry is not necessarily the one that does not close. It is the one that requires many rounds of basic education before anyone learns that the property, budget or timing was never workable.
Sending an agent twenty unrelated links is not a shortlist. A useful shortlist records why each property is there and what would make the buyer reject it.
Most intake forms capture contact details but little reasoning. An AI conversation can capture how the buyer arrived at a shortlist, which constraints are hard and which are flexible.
Cross-border real estate often involves more than one organization: the discovery platform, the buyer-side coordinator and the licensed local firm with the listing or transaction authority. Information quality at the handoff determines how much work is duplicated.
The translation problem is not only Japanese to English. It is market context: why the 1981 seismic cutoff matters, why assessed value differs from asking price, why station distance is not commute time and why 'as-is' is not an inspection.
An agentic search product should have a designed stopping point. For Engawa, that point is when the shortlist exists but current availability, documents, viewings, seller communication or transaction steps are needed.
Bilingual agents often become the translation layer, market educator, search assistant and transaction coordinator at once. An MCP-enabled research layer can offload the repetitive first three without replacing the last one.
Rural inventory has the largest gap between what can be digitized and what matters on the ground. That makes it a good fit for AI discovery and a bad fit for AI-only transactions.
Large portals can absorb the cost of multilingual product teams. A small brokerage usually cannot. MCP creates another route: let a shared discovery layer handle structured buyer research, then let the local firm handle the work only it can do.
Lead qualification is a loaded term. For an AI property tool, the safe and useful part is search qualification: does the buyer have a budget, target, timeline and shortlist that makes sense?
Search engines, portals and social media are all distribution channels. MCP adds a new one: the buyer's AI assistant can call a property tool directly while solving a broader task.
AI discovery is not one channel. Models can answer from indexed web content, retrieved pages, structured feeds or live tools. MCP is the live-tool path: the assistant asks the server for current structured data when the user needs it.
The useful question is not whether AI replaces the agent as a job title. It is which tasks become cheap enough to automate and which tasks become more valuable because the buyer arrives better informed.
A warm lead is one with evidence of a decision process. A user who has compared four listings and asked for the assessed value of one is different from a user who downloaded a generic guide.
The control question is permission design. A read-only tool that helps a buyer research is fundamentally different from an agent that can email sellers, submit offers or alter CRM records.
Foreign buyers need a digital layer because they begin far from the property and often outside the Japanese-language market. Local agents need a professional layer because the decisive information and transaction work cannot all be digitized.
GEO is not a replacement for SEO. It is the discipline of making your information usable when an AI system is answering a buyer's question rather than merely ranking ten blue links.
There is no single switch that guarantees an AI citation. Different systems use search indexes, retrieval, browsing and tools. The common requirement is that the information can be found and interpreted reliably.
Schema.org markup is useful for the web, but agent-ready data is broader. The AI also needs tool inputs, output semantics and guidance about which fields can be compared safely.
SEO is still the foundation for public discovery, brand authority and crawlable information. MCP does not replace it because a tool connection is not the same thing as a searchable public corpus.
SEO asks whether a page can be crawled, understood and ranked for a query. GEO adds another question: if an AI retrieves this information, can it extract a clear, attributed answer from it?
A language model can discuss general real-estate concepts from its training, but a question such as 'what houses under ¥5M are available in Chiba?' requires current external data.
Real estate decisions are unusually sensitive to provenance because two identical-looking numbers can have very different meanings. An official hazard reading, a seller description and an AI estimate should never be flattened into one confidence level.
An AI interface compresses information. Compression is useful, but it can separate a fact from the context that made it trustworthy. Source links reconnect the summary to evidence.
The web created a distribution layer of pages and links. Mobile created apps. MCP creates a tool layer where the user's AI application can call external capabilities while solving a task.
The near-term change is not autonomous house buying. It is autonomous research: an assistant can search, inspect, compare, calculate and keep the buyer's constraints across a long decision process.
Choose Engawa when curation, AI-assisted shortlist work and a smaller guided workflow matter more; choose AkiyaMart when broad map-led inventory and a mature discovery-to-purchase pipeline are the priority.
Choose Engawa for a curated shortlist plus coordination; choose Akiya Japan for the lowest-cost broad akiya search and self-directed research.
Choose Engawa for rural, akiya and curated buyer support; choose Real Estate Japan for large conventional urban inventory, rentals and market learning.
Choose Engawa when buying a house is the main project; choose wagaya Japan when buying, renting and relocation need to live in the same foreigner-oriented search workflow.
Choose Engawa for a smaller, guided shortlist; choose Japan-Property.jp for a conventional multilingual portal spanning more property categories.
Choose AkiyaMart if you want richer map and hazard-oriented discovery plus a clearly packaged $5,000 guided-purchase route; choose Akiya Japan if you want a much cheaper full-access search subscription and are comfortable assembling support separately.
Choose AkiyaMart for akiya-led nationwide discovery and remote-buying support; choose Real Estate Japan for mainstream urban, investment and rental inventory with a long-running English portal.
Choose Akiya Japan for inexpensive nationwide akiya discovery; choose Real Estate Japan for large conventional for-sale inventory, rentals and English market education.
Choose wagaya Japan when foreigner-friendly renting, monthly housing and buying all matter; choose Real Estate Japan when sale inventory, investment property and a conventional English property portal are the main needs.
Choose Japan-Property.jp if you prefer a multilingual portal spanning residential, land and commercial categories; choose Real Estate Japan if you want a larger English ecosystem with rentals, seminars and a dedicated Property Finder.
Engawa is the closest alternative when you want guided curation; Akiya Japan is the clearest lower-cost alternative for broad akiya search; Real Estate Japan is better when your search has moved into mainstream urban or investment property.
AkiyaMart is the strongest alternative for a larger packaged discovery-to-buying workflow, while Engawa is better for a curated shortlist and Real Estate Japan is better for mainstream urban property.
wagaya Japan and Japan-Property.jp are the closest conventional portal alternatives; AkiyaMart and Akiya Japan are better for akiya; Engawa is better when the bottleneck is shortlisting and English coordination rather than inventory volume.
Use AkiyaMart or Akiya Japan for broad English akiya discovery, municipal akiya banks for source-level local inventory, and Engawa when you want a smaller shortlist with context and buyer coordination.
For mainstream property use Real Estate Japan, wagaya Japan or Japan-Property.jp; for akiya use AkiyaMart or Akiya Japan; for a curated shortlist and coordination use Engawa.
Real Estate Japan is the strongest general English portal, wagaya Japan is particularly useful for foreigner-oriented rent-and-buy searches, and Japan-Property.jp adds another multilingual source across residential and commercial categories.
For maximum cheap-house discovery use AkiyaMart, Akiya Japan and municipal akiya banks. Use Engawa or a local agent when you need to narrow the list based on access, condition, hazards and transaction practicality.
Use municipal akiya banks for local source inventory and programme details, and use commercial portals for faster cross-region search, filters and market comparison.
Yes, a foreign national can generally buy a house and land in Japan. Buying property does not itself create a visa or right to live in Japan.
Non-residents can generally purchase Japanese real estate, but should design the transaction around remote identity verification, banking, tax administration and representation from the start.
Remote buying is possible when the seller, handling agent and judicial scrivener can support it and the buyer supplies acceptable identity documents and authority for local acts.
Residents may use Japanese resident records and seal/signature documents; non-residents commonly need foreign official or notarized substitutes plus translations where required.
Permanent residents with stable Japanese income have the widest options. Other long-term residents may qualify at some banks; buyers living abroad should expect a much smaller lender set, larger equity requirements or cash financing.
Cash is often the practical route for non-residents and low-cost akiya, but 'cash' usually means bank funds transferred at settlement, not physical currency.
Budget separately for one-off acquisition taxes and recurring ownership taxes. Do not estimate every charge as a simple percentage of the asking price because several taxes use assessed value instead.
Expect annual fixed-asset tax and, where applicable, city planning tax, plus separate income-tax obligations if the property produces rent.
The Ministry of Finance states that a non-resident who acquires real property in Japan or rights in it must generally submit a report through the Bank of Japan within 20 days after acquisition, subject to the current rules and exceptions.
Foreign buyers use the same real-property registration system. The difference is usually the evidence used to prove identity and address when the buyer does not have ordinary Japanese resident documents.
For a foreign buyer, the judicial scrivener is especially important because they confirm what identity and address evidence can support registration and whether a power of attorney is acceptable.
Build a line-item estimate before making an offer. The important inputs are assessed value, broker fee structure, registration, one-off acquisition tax, professional fees, stamp duty and insurance.
A ¥2 million house can be a bargain or an expensive liability. The difference is usually roof and structure, water/sewage, access, renovation labor and whether the building is legally useful after the work.
Do not let a low price compress diligence. The cheaper the house, the more likely fixed repair and legal problems dominate the economics.
Use a building professional to inspect the property before you become contractually committed whenever the transaction allows it, and add specialist checks when the building age or site makes them necessary.
A pre-1981 house is not automatically unsafe and a post-1981 house is not automatically problem-free, but the cutoff is a major diligence signal and can affect financing, insurance, resale and renovation decisions.
A post-2000 timber house is not automatically superior in every respect, but the date is useful when comparing older wooden homes and deciding how much structural investigation to commission.
Before buying, confirm which legally recognized road serves the site, whether required frontage exists, and whether private-road rights or setbacks affect the property.
Do not assume ownership of a house includes an automatic right to rebuild it. Confirm current building-law access, zoning and site restrictions with the municipality or a qualified professional.
Yes, foreigners can generally buy and register Japanese land in their own name, including as non-residents. Buying land does not grant a visa or permission to live in Japan.
Start by separating legal ownership from programme eligibility. You may be allowed to own a property even if a municipal akiya-bank subsidy or matching programme has residency conditions.
Foreigners can generally buy kominka under the same ownership rules as other houses. The challenge is condition, specialist renovation and making a traditional building work for your intended use.
Foreigners can generally buy rural houses, including land. The key question is whether the location works as a place to live or manage, not whether the house is inexpensive.
Treat ¥5 million as a search ceiling, not a total-budget ceiling. A viable project may need the same amount again—or much more—in repairs, taxes, fees and setup.
A ¥10 million ceiling is often more useful than a ¥3 million ceiling because it lets you buy condition and location instead of spending the difference later on repairs.
Shortlist prefectures by lifestyle and operating constraints first, then compare towns. Prefecture-level price averages are less useful than the difference between a connected regional city and a remote village.
Choose Tokyo for the deepest market and global demand; choose Osaka when you want a major-city property at a generally lower entry point and are comfortable with a somewhat smaller market.
Foreigners can generally buy Hokkaido property like elsewhere in Japan. Buyers should focus on winter access, heating systems, snow load, frozen-pipe risk and who will manage the building when vacant.
Foreigners can generally buy Okinawa property. The main diligence questions are building durability, access and whether the specific island or municipality fits your intended use.
Foreigners can generally buy Nagano property. Resort areas can have international demand and higher prices; less-touristic towns can offer cheaper houses but thinner services and resale.
Do not confuse legal accessibility with investment quality. Japan is open to foreign ownership, but non-resident tax, financing and management can make a theoretically high-yield property unattractive in practice.
Calculate yield on total capital invested, not asking price. Then stress-test rent and occupancy using local comparables and a realistic maintenance reserve.
Flips, yield, land banking and long stays. Four plays, and the exact prompts that make an AI assistant do the search, the shortlist and the closing-cost maths in one conversation.
Six tools, one API key, and a design rule that matters more than any of them: an unknown is written out in words rather than left as a missing field.
There is no citizenship or residency requirement to own Japanese property. The real constraints are financing, tax residency and paperwork, not your passport.
Closing costs add roughly 6–10% on top of the asking price, and renovation frequently exceeds the purchase price. Here is the whole number, line by line.
A practical comparison of six active English-language services, including broad portals, akiya databases, and curated buyer support.