Seoul-based travel data company DealLabs announced on October 28, 2026, that it has built a database of verified local travel information for seven Asian cities, containing more than 280,000 records as of October 13, each accompanied by its source and verification date. The database powers more than 23,000 guide pages and is published in formats AI systems can read directly, in a bid to make travel answers from generative AI accurate for the city and the day in question.
What The Database Covers
The seven cities are Seoul, Busan, Tokyo, Osaka, Bangkok, Taipei and Singapore. The database covers what travellers look for after arriving: subway station exits, timetables, first and last trains, fares, restaurants, tourist attractions, hospitals, closing days for Singapore’s hawker centres, and business hours for Taipei’s night markets. It is, in effect, an attempt to turn the messy, fast-changing details of a city into a structured layer that an AI assistant can query rather than guess at.
The scale varies sharply by city. In Osaka alone, the database contains more than 114,000 records, including more than 42,000 medical institutions and pharmacies and more than 30,000 restaurant permits. In Taipei, it contains more than 63,000 records, including 7,200 station-to-station fares. Those figures show where the work is heaviest: not in landmarks, but in the regulated, frequently updated records of health, food and transport that travellers actually depend on.
All seven cities have AI-ready guide files and data APIs, while Seoul also provides a standard API specification. The seven cities run on a single common engine, DealLabs says, so a new city can be added simply by incorporating that city’s data and sources. DealLabs plans to expand coverage to 10 cities by the end of 2026 and to 30 cities, including New York and Paris, by 2028.
Why Travel Answers Go Wrong
Travel information changes rapidly, and DealLabs’ announcement leans on that problem. On a single day, public event listings and organisers’ own notices gave different end dates for three of 25 pop-up stores in Seoul, up to ten days apart. Business hours, closing days, last train times and fares are among the areas where AI travel answers are most frequently inaccurate, the company says, and it continuously updates its information using official sources.
That framing matters because generative AI travel answers sound confident even when they are stale. A chatbot that invents a closing time or a fare does not look uncertain; it looks helpful, and the traveller discovers the error at a closed restaurant or a missed last train. A verified data layer does not remove that risk entirely, but it changes the failure mode: the answer can be traced to a source and a verification date, rather than to a model’s best guess.
DealLabs is also piloting Visit Plan Check, which checks travel AI itineraries against official sources daily. The idea is to catch drift after an itinerary is made, not just at the moment it is generated, since a plan made weeks ahead can quietly become wrong as hours, dates and fares change.
A Surge In AI Demand For The Data
Demand for the data is rising fast, according to DealLabs. The number of times generative AI search and answer services accessed its data increased from around 600 times per day on a seven-day average in late September to more than 11,000 times per day in mid-October, nearly twentyfold in less than a month. That surge suggests AI services themselves are looking for grounded local data, not just generating answers from their training data.
Founder and chief executive Young Cho put the point simply: when travellers ask AI a question, what matters is whether the answer is accurate for that city today. DealLabs says it aims to establish a standard for local data that travel AI can rely on, a claim that will be tested by how widely AI providers actually adopt the layer, and by how fresh the records stay as the company adds cities.
The announcement does not disclose pricing, licensing terms or which AI services are already using the data beyond the access figures quoted. Those details will determine whether this becomes a shared infrastructure layer for travel AI or a proprietary dataset used by a few partners.
Why It Matters
This matters because AI travel planning is moving from novelty to infrastructure, and its weakest link is local accuracy. A verified, sourced, dated layer for station exits, fares and opening hours is less glamorous than a new model, but it is the kind of data work that decides whether an AI itinerary is useful or misleading. The same lesson runs through other AI infrastructure stories, from a single source of truth for AI factories to clouds opening to rival chips: as AI spreads, the winners may be the companies that own the reliable layer underneath. If DealLabs can keep 280,000 records fresh across seven cities and scale to 30, it will have built something AI assistants genuinely need, in the same way interactive AI worlds depend on the data behind the experience.
Background
DealLabs is a Seoul-based travel data company that builds verified local information for travellers and AI systems. Its database is published in AI-readable formats, including guide files and data APIs, and it says it updates its records continuously from official sources.
Source: DealLabs announcement
0 Comments