Tamil-language chatbot
Saaral Travel
"திருவனந்தபுரம்" resolves to the canonical city in milliseconds. Non-Latin scripts stop blocking checkout.
find_city
The travel data layer for AI products that can't afford hallucinated geography. 37,685 cities · 4.7M POIs · 43 grounded tools · license tag on every answer.
8 stories from the use-cases page — each tile is a real-shaped travel business and the API call that powers it.
"திருவனந்தபுரம்" resolves to the canonical city in milliseconds. Non-Latin scripts stop blocking checkout.
4-second destination dossier replaces a 40-minute manual brief. 3× more prospects quoted per day.
New "Coastal Karnataka" line ships in a day instead of a quarter — one circuit call, members in order.
Vangchhia necropolis surfaces alongside its better-known peers — invisible-in-guidebooks heritage made browsable.
Central, Maido, Kjolle ranked Michelin-tier-first. When the guide updates, the tour app updates with it.
Treasury entry → next-screen offers Monastery hike + canyon restaurant + viewpoint. AOV up, no rec engine.
The seasonality curve for Reykjavík becomes the page that ranks, converts, and books.
Government PDFs become a chat-answerable knowledge base overnight — every result links back with a page number.
Each stamp is one Concierge tool — country postmark, function name, what it returns.
Showing 18 of 43 · see all tools mapped to use cases →
Every row carries its source URL, license tag, and attribution requirement. Compliance reviews close in minutes, not weeks.
Master score across 37,685 cities covering safety, sentiment, accessibility, infrastructure.
Sub-200ms p99 reads. 23M-node travel graph for proximity, routing, similarity, themed clusters.
Every entity in the prose appears in the structured layer; citations link back to source.
Honest comparisons against the obvious comparables — not feature checkboxes, just where each one wins and loses for AI-grounded travel use cases.
Places is for showing a pin on a map; we're for grounding a paragraph. Different stack, different price model.
Sabre wins on live availability and GDS bookings; we win on narrative grounding and content licensing.
OSM is free; the cost is curation, license tagging, and the months you'll spend re-deriving city names from POI clusters.
Wikipedia is great; scraping it at scale puts you in licensing territory you don't want to learn about by accident.