DESTINATION INTELLIGENCE INFRASTRUCTURE

AI is becoming the interface for travel.
It has no source of truth.

Tabibito is the AI-native Destination Intelligence Layer™ - structured, curated, verified
knowledge about places, consumed by assistants, apps and platforms.

Request access See how it works

Models can reason. They can't know.

01

What LLMs are good at

Language, inference, summarising, planning. They will keep getting better at all of it.

02

What they have no source for

Whether it's open on a Tuesday. Whether it takes your card. Whether it's genuinely gluten-free in a country where soy sauce contains wheat.

03

What Tabibito supplies

Structured, curated, verified destination knowledge, as a graph rather than a list. Consumed by any interface.

Every travel assistant on the market today is reasoning fluently over a knowledge base nobody built.

Same question. Two knowledge bases.

Illustrative. This example shows where the layer is going. Opening hours and closure days are live in the dataset today; walking times, prices and dietary attributes are on the build order and are not yet held. The query below this runs against the real graph.
QUERY

"We're in Tokyo near Ueno with two kids under ten. It's Tuesday, it's raining, we have three hours before dinner, about ¥5,000, and one of us is coeliac."

Generic AI assistant

Fluent. Plausible. Unusable.

Since it's raining, you could visit the Tokyo National Museum or Ueno Zoo. For gluten-free dining, look for restaurants offering sushi or rice-based dishes, which are naturally gluten-free. Many places in Tokyo now cater to dietary restrictions - it's a good idea to check reviews or ask staff.

  • ✕Doesn't check Tuesday hours
  • ✕Sushi is not reliably gluten-free in Japan - soy sauce is wheat-based. This advice is unsafe.
  • ✕No transit time, no cost, no card acceptance
  • ✕"Check reviews or ask staff" hands the work back to the traveller

Tabibito

Specific. Sourced. Actionable.

National Museum of Nature and Science - 6 min walk, indoor, open till 17:00 Tue, ¥630 adults / under-18 free. Hands-on floor built for this age.

Then: A soba restaurant nearby - 11 min, confirmed gluten-free tamari kitchen, staff briefed, kids' seating. ¥3,200 for four.

Avoid the Ameyoko food stalls - cash only, no allergen labelling, covered but not dry.

  • ✓Opening hours verified against Tuesday, not assumed
  • ✓Gluten-free means tamari kitchen confirmed, not "probably fine"
  • ✓Fits three hours and ¥5,000 with change
  • ✓Human-curated, timestamped, improving with use
LIVE QUERY

Ask a general model what is open today. Then ask us.

237 of our 1,884 records carry an explicit weekly closure rule. 89 shut on Mondays. A model answering from memory has no way to know that, and will recommend them anyway.

In Tokyo on a Tuesday, 17 places we hold are shut and 260 are open. We do not know for 12, and we say so.

CLOSED THAT DAY
  • ✕Sosakumen Kobo Nakiryu · Toshima · Ramen Shop11:00am to 3:30pm
  • ✕Yakumo · Meguro · Ramen Shop11:30 am to 3:30pm
  • ✕Corn Barley · Musashino · Anime11:00am to 10:00pm
  • ✕Ladrio · Chiyoda · Cool PlaceWeekdays- 11:30am to 10:30pm
  • ✕Gavial · Chiyoda · RestaurantMon- 11:00am to 4:00pm
  • ✕Ghibili Museum · Mitaka · Experience Centre10:00am to 6:00pm (Closed on Tuesday)
OPEN INSTEAD, HIDDEN GEMS FIRST
  • ✓Tatsu no ya · Shinjuku · Ramen Shop11:00am to 10:00pm
  • ✓Fuunji Shinjuku · Shibuya · Ramen Shop11:00am to 3:00pm &
  • ✓Ramen Kamo to Negi · Taito · Ramen ShopOpen 24 Hours
  • ✓Ramenya Shima · Shibuya · Ramen Shop8:45am to 2:00pm
  • ✓Shunya-chan · Hachioji · Ramen Shop6:00pm to 11:00pm
  • ✓Ichiran Ramen · Shinjuku · Ramen Shop12:00pm to 6:00am &

Computed live from the graph on every change. Opening hours are parsed from source text supplied by the people who visited; 1,761 of 1,884 records parsed with high confidence, and the rest are reported as unknown rather than guessed.

One record. Not a row - a node.

The moat isn't how many places we hold. It's how much each one knows about everything around it.

20,666
in the same area
near
15,292
similar in kind
similar
1,497
open when this one is shut
alternative_when_closed
1,492
found by the same creator
same_creator
638
part of the same curated day
in_bundle

39,585 relationships across 1,884 records. Counted from the graph as it stands, not estimated.

AND WHAT EACH RECORD HOLDS
1,884
editorial description
1,873
ward or area resolved
1,789
opening hours parsed
237
explicit weekly closure rule
791
flagged as a hidden gem
1,133
creator or social provenance
1,881
map reference
1,884
persona tags

Every figure on this page is counted from the graph, never typed in, and refreshed hourly. Fields still being populated, including coordinates, dietary handling and payment acceptance, appear here on their own as they land.

We expose intelligence, never raw tables.

// not this
GET /restaurants?city=tokyo&type=ramen
// this
POST /v1/recommend
{
city: "Tokyo", near: "Ueno",
party: "2 adults, 2 under-10",
constraints: ["coeliac", "indoor"],
window_hrs: 3
}
→ reasoned · ranked · constraint-aware · with provenance
MCP TOOLS
search_recommendationsPLANNED
destination_briefingPLANNED
build_itineraryPLANNED
nearby_discoveryPLANNED
neighbourhood_comparePLANNED
seasonal_recommendationsPLANNED
budget_planPLANNED
merchant_intelligencePLANNED

One graph. Every interface.

Trundle

SHIPPED

Our consumer app, live on Google Play and the App Store. The first client of the layer, and how we keep the graph current.

MCP server

Any AI assistant calls Tabibito as a tool.

Enterprise API

Airlines, hotels, tourism boards and travel platforms embed the same layer.

Infrastructure, not an app.

Travellers
Trundle, other consumer apps, assistants they already use
AI assistants
Models that reason well and cannot verify
Travel companies
Airlines, hotels, OTAs, tourism boards, corporate travel
TABIBITO
The destination intelligence layer
Supply and transaction rails
Flights, hotels, visas, eSIM, insurance, payments

We deliberately don't own the layer below. It has incumbents, thin margins and no differentiation.

Four countries. Deeply.

We build destination by destination, not country by country. Japan, South Korea, Thailand and Vietnam are live in the graph today. We would rather be the definitive source for a few countries than a thin index of a hundred.

JAPAN - LIVE SOUTH KOREA - LIVE THAILAND - LIVE VIETNAM - LIVE

Building something that needs to know where it's sending people?

We're onboarding a small number of design partners - AI products, travel platforms and tourism boards - while the API firms up.