TravelQuoteBot

Tour operators build quotes by hand every day — tracking hotel, tour, transfer, guide, museum and meal prices in spreadsheets, sending the customer a 5-15 page PDF, then re-doing everything when a price changes. With 50+ line items a single quote takes 2-3 hours.

The result: stale-price risk, weak personalization, copy-paste templates, less customer-facing time. Operators end up managing data instead of selling trips.

Multi-tenant SaaS architecture (organization_id isolation). Every operator owns their pricing catalog, templates and customers. Broad pricing modules: hotels, tours, vehicles, yachts, transfers, flights, meals, museum entries, guides — all category-aware (pax count, season, room type).

An AI itinerary generator (Anthropic Claude SDK): customer brief → personalized itinerary plus pricing pulled from the operator's own catalog, in seconds rather than minutes. Branded react-pdf output; subscription engine (14-day trial → monthly), super_admin impersonation, customer-facing preview links.

Active Google + Meta campaigns run 24/7 for operator acquisition — which is why TQB ships only at night. Cost-aware architecture: when Google Maps got expensive we swapped to OpenStreetMap, kept the code path, kept the margin.

30 organizations total, 5 active paid subscriptions, 27 trials. Aggregated inventory: 2,018 hotels, 1,708 suppliers, 544 tours, 331 guides, 309 vehicles spanning 29 countries. 110 AI itineraries generated; 245 activity-log entries.

The anchor-tenant thesis (drink your own champagne) held: in-house Funny Tourism is the most active user with 107 events; the top two operators carry 54% of all activity. The Q4 2025 launch wave added 11 new operators in a single month.

Replicable insight: improving trial-to-paid conversion (currently 16.7%) is higher-ROI than acquiring new trials. Onboarding friction analysis + activation-event design is the next move.