YURAGI Travel Releases an MCP Server Built for Luxury Inbound Travel to Japan

YURAGI, the inbound travel service operated by Willverse Inc. (headquartered in Tokyo, Representative Director Haruto Asahina), is releasing an MCP server that lets the AI agents used by overseas travel agencies query Japan-side experiences, guides, and itinerary-design criteria directly. MCP (Model Context Protocol) is a standard that allows AI to call external data and services.

When an agency's staff tells their AI "seven days including Kyoto and Kanazawa, a family of four, interested in tea ceremony and crafts," the AI queries YURAGI and retrieves bookable experiences and their availability, the conditions of guides who can take the assignment, and the itinerary-design criteria YURAGI has accumulated in its vector database. The agency's AI assembles the itinerary; YURAGI supplies the criteria and handles the actual arrangements. An itinerary at a level of precision a general-purpose LLM alone cannot produce comes together on the spot.

How This Differs From Asking a General-Purpose AI

Ask ChatGPT, Claude, or Gemini to "build a seven-day itinerary including Kyoto and Kanazawa" and you will get a reasonable answer. Well-known destinations, arranged in a sensible order. There are three differences.

Whether it can actually be booked. The itinerary a general-purpose AI returns is a plausible-looking plan assembled from public information. It cannot tell you whether that experience has availability on the day, whether the minimum party size is met, or whether the venue is closed. It is not unusual for the whole plan to be rebuilt once arrangements begin. YURAGI's server queries inventory that is actually bookable.

Whether the information satisfies luxury travelers. What exists on the web is records and reviews of trips that went well, aimed at a mass audience. Things you only learn after getting them wrong — "visiting in this order leaves guests tired by the afternoon," "a transfer that looks short on a map but isn't realistic on the ground" — are not published, and once you narrow to luxury travel, there is little on the web at all. What we have accumulated is a record of these small burdens: unstructured data specific to luxury travel. A general-purpose AI's knowledge is the average of public information, and what determines the quality of a trip lies in the part that deviates from the average. However capable the models become, information written down nowhere will not surface.

Whether someone will execute it. An itinerary is still only paper at the moment it is written. On the day, the guide and the chauffeur arrive, the car moves, and when the plan breaks it has to be rebuilt. That requires a party who takes responsibility. On YURAGI's server, a completed itinerary is filed directly as an arrangement request and routed through to a YURAGI staff member. The AI supplies the criteria; people hold execution and responsibility.

Proposals That Connect to a Customer's Past Trips

Where this server proves its worth in practice is when it is combined with the customer information an agency already holds.

Suppose a customer who toured temples and gardens in Kyoto three years ago comes back to discuss a return trip. The agency's AI reads the past itinerary from its own customer management system and queries YURAGI. The same Kyoto, but composed without repeating the previous visit; craft and culinary experiences the customer had little interest in at the time; a guide with a different specialty than before. All of it can be assembled from inventory that is actually bookable.

Proposals of the form "they went here last time, so this time we do this" have until now been built through a back-and-forth in which staff queried the local operator, waited for a reply, and asked again. That back-and-forth is replaced by a single AI query.

In this exchange, YURAGI does not receive customer information. Queries can be made without passing names or contact details, and customer data stays on the agency's side. By design, only the Japan-side criteria flow across.

Itinerary Design Has Been a Black Box

As inbound demand has grown, the role of the DMC handling local arrangements has become larger year by year. At the same time, this work retains structurally opaque areas.

Why an itinerary took the shape it did, and what a price is composed of. These are hard to see from the ordering agency's side, and as a result business proceeds without quality being comparable. The differences plainly exist, yet the language for explaining them collapses into "the experience of the person in charge." As a service industry, we see this as an area with substantial room for improvement.

Through handling high-value tailor-made travel, YURAGI has continually tested the precision of its itinerary design and the soundness of its quotes. Even for the same "three days in Kyoto," the itinerary that should be offered to a family visiting Japan for the first time differs from the one for a traveler on their fifth visit who has seen enough temples. What determines that difference is not the volume of knowledge but the accumulation of how proposals were made in the past and where they were corrected.

This server returns the criteria extracted from that accumulation. For example: "to avoid crowds, place the Arashiyama bamboo grove first in the day, arriving before 9 a.m.," or "assign guides to cities with high cultural density and make transfer days self-guided" — material at the level of what was learned by actually doing it.

Where Value Moves in the Age of AI

The information AI can handle divides broadly into two kinds: structured data retrievable from databases and APIs, and unstructured data that exists as text and records and only becomes searchable once it is vectorized.

In the travel industry, only the former has been structured. Inventory, availability, price. These circulate via API and anyone can obtain them. The more AI agents spread, the more this layer is compared side by side and the less it differentiates.

Almost everything that determines value in luxury tailor-made travel sits on the latter side. Why visit in that order. Why assign a guide on this day and not another. Where the previous proposal with this composition missed. These live in proposal documents, emails, and the memory of the person in charge — not merely unstructured, but never even consolidated as text.

Vector search and RAG (retrieval-augmented generation) are themselves already commodities. With the available infrastructure, anyone can stand one up in a short time. What matters is only whether you hold the substance to put into it.

And while unstructured data on luxury travel is scattered across operators, to our knowledge no operator has yet established it in a form AI can read. In this area, effectively everyone is still standing at the same starting line. Somewhere other than the performance race between general-purpose models, an untouched contest remains.

Organizing the layers of travel arrangement by the nature of their data yields four. The first is inventory, availability, and price: structured data that sits in reservation systems and circulates via API. Anyone can obtain it, so it does not differentiate. The second is reach into supply (guides and local relationships): semi-structured data that depends on contracts and relationships specific to each operator. It is relationships rather than data, which makes it hard to copy.

The third is the intent behind a proposal and its outcome. This is unstructured data, consolidated nowhere. Only whoever vectorizes it can turn it into an asset. The fourth is execution and responsibility on the day. It cannot be turned into data at all and depends on the capability of the local operator, so AI will not replace it.

What YURAGI took on was to accumulate the third layer — the intent behind a proposal and its outcome — on a vector search platform, and to bundle it with the first layer (inventory, availability, and price) so that both are returned through the same interface. The record of how proposals were made and where they were later corrected is the source of the criteria passed to the agency's AI. In a single query, the agent obtains both "what can be booked" and "why it should be composed this way."

Where Travel Agencies and DMCs Compete From Here

Structured inventory data is now consolidated in OTAs such as Viator, GetYourGuide, and Booking.com. Traditionally, part of a travel agency's value lay in accessing that information and organizing it on the customer's behalf — the work of closing an information asymmetry.

That part thins out as AI agents become able to pull OTA data directly. In fact, major OTAs have already begun measuring bookings that arrive via AI.

So where should the contest be fought? We believe the winning path is to accumulate unstructured data on the customer side — who prefers what, what they tire of, how they have changed — and to move to the side that proposes, as a concierge. This information does not collect in OTAs. It accumulates only in the agencies and DMCs that hold a direct relationship with the customer.

In the age of AI, what each player should accumulate diverges. OTAs accumulate structured inventory and pricing. Travel agencies and DMCs accumulate the customer's Knowledge Layer: preferences, history, and change. And the local arrangement company — where YURAGI Travel stands — accumulates local supply, execution, and the criteria behind proposals.

YURAGI is releasing this MCP server on the premise of that division of labor. The agency holds its own understanding of the customer; YURAGI holds the Japan side. Only when both are in place can a returning traveler be told "you went here last time, so this time we do this." YURAGI is designed not as a competitor to agencies but as the connection point that carries the Japan side.

Looking Ahead

The service currently targets agencies and DMCs in English-speaking markets. Going forward, we will progressively expand supported languages and responses aligned with local business practices for VIP markets across Asia, including Taiwan, South Korea, and Indonesia.

The high-spend segment of travel to Japan is growing not only in the West but across Asia, and the way an itinerary should be composed differs by market. While connecting with partners in each market, we will grow the criteria YURAGI returns on a market-by-market basis.

Contact

YURAGI works with domestic travel agencies, DMCs, and local content operators to produce high-value experiences for inbound travelers. If your company is interested in attracting international travelers, we would welcome the conversation. (Email: yuragi-support@willverse.io)