A travel OS for discovery,one-click booking, trip management, andlive AI copilot execution.
- Type
- Web app, AI travel platform
- Category
- AI & LLM
- Stack
- Next.js / React, TypeScript & Tailwind, Node & Express
- Live site
- dev-voyiage-ai.vercel.app
The challenge
Planning a trip normally means juggling separate tools for inspiration, itineraries, bookings and day-of changes. Voyaige AI keeps the whole journey in one place, and bookings are never auto-executed: each one needs an explicit confirm step.
Voyaige AI brings travel planning, booking and trip management into one product. A chat learns the traveller's style and builds itineraries, activities, cars, flights and stays can be booked in one click, and a live copilot handles dining, routing, weather and sold-out alternatives. The product is in development (beta) and stable.
What it does
- 01
AI planning chat
A streaming chat learns the traveller's style, using OpenAI embeddings and pgvector for preference retrieval, and builds itineraries on a map.
- 02
One-click booking
Activities, cars, flights and stays go through basket and checkout with traveler details, a summary, Stripe payment and confirmation.
- 03
Trip organizer
Trips are organized into days and activities, with a list view and a detail page for each trip.
- 04
Live AI copilot
The copilot suggests dining, routing, weather-aware changes, ideas for free hours and alternatives when something sells out.
- 05
Secure architecture
A modular backend of 22 modules sits behind Next.js Server Actions and proxy routes, so the browser never calls it directly, and tokens live in an encrypted httpOnly cookie.
Built with
- Next.js / React
- TypeScript & Tailwind
- Node & Express
- Prisma & PostgreSQL
- Redis
- Anthropic Claude & OpenAI
- Stripe
- AWS
At a glance
- 4
- Product pillars: planning, booking, trip management and a live copilot
- 22
- Modules in the modular backend
- SSE
- Chat streamed live to the browser
- Confirm
- Every booking needs an explicit confirmation step
