TravelMindsAI is a grounding layer for travel AI: curated cities, points of interest, lodging, visa rules, UNESCO sites, and composite destination scores — joined to Wikidata and Wikipedia for explainable answers. Designed for retrieval, not as a thin LLM wrapper.
All numbers reflect the current curated data layer. Coverage is updated as ingestion pipelines run.
Deduplicated, ISO-coded, lat/lon-resolved.
Foursquare-derived; duplicates and tribute pages excluded.
Cross-source deduplicated, license-tagged.
Including Michelin-licensed entries.
Tours and activities, deduplicated.
Including transboundary sites.
Plus per-passport openness scores for 199 passports.
From Wikidata under permissive licenses.
A retrieval substrate, not a chatbot.
Every row carries its source, license, and a license-passed flag. Commercial-public buckets are separated from internal-only at the schema level.
NULL-aware master score blending social, connectivity, sentiment, and value-for-money signals — with explicit coverage penalties.
Postgres + pgvector for app reads; OpenSearch for hybrid search; Neo4j for graph context; DuckDB for analytics.
City and POI narratives are generated from the curated data with hallucination checks against an explicit context manifest.
TravelMindsAI is in early access. We are not yet listed on third-party API marketplaces. Pricing and refund terms are published so that early-access customers can evaluate the platform with full clarity. The order process for paid plans is conducted by our online reseller Paddle.com — Paddle is the Merchant of Record for all our orders, and handles customer service inquiries and returns.