Conversational AI travel platform
Voxia — AI Smart Travel Assistant
01 / Problem
Business travel planning is fragmented: searching flights and hotels, generating travel-request paperwork, and filling out PDF forms all happen in different tools with no shared context. The goal was a single assistant that understands a travel request in natural language and carries it through to the paperwork.
02 / Approach
Built the conversational layer on Rasa for intent recognition and dialogue management, backed by a Node.js/Express API that orchestrates flight and hotel search, travel-request generation, and PDF auto-filling. The React frontend gives the assistant a real interface rather than a chat-only widget. Python services handle the document-automation side, parsing and populating PDF forms from structured trip data.
03 / Architecture
Rasa core for NLU/dialogue, an Express API gateway coordinating search and document services, MongoDB for conversation and trip-request state, and Python workers for PDF generation. Deployed on Render with Docker, fronted by Cloudflare.
04 / Outcome
A working assistant that takes a travel request from natural-language input through to filled paperwork, proving conversational AI could carry real document-automation work, not just answer questions.