Live AI voice agent that builds complete trip itineraries through natural conversation.
Overview
Tango runs large-scale university travel across the US — a business that traditionally lives in spreadsheets, email threads, and manual reconciliation. We built them a complex, 4-sided marketplace from scratch that manages everything from charter flight inventory to automated payment routing and seasonal pricing logic, replacing manual coordination with a single, intelligent platform. Its standout capability is a live AI voice agent: travelers now plan an entire trip just by talking, and a complete, priced itinerary assembles itself in real time — turning what used to be days of back-and-forth into a single conversation.
The Challenge
Organizing university tours is a logistical puzzle involving four distinct personas: Students, University Heads, Vendors (flights, hotels, catering), and Internal Ops teams. Each has different needs, permissions, and workflows. Tango needed a system that could handle nested inventory and seasonal pricing volatility — cancellations, partial refunds, and fluctuating vendor rates — without constant manual intervention.
The Engineering Journey
We built the entire platform from the ground up: FastAPI for a robust, high-performance backend and Next.js for a fast, mobile-friendly frontend. The centerpiece was a Rules Engine designed to handle the hundreds of what-if scenarios inherent to travel. Rather than hard-coding logic, we architected it as a composable rule graph — so Tango's ops team can add new pricing or refund rules without an engineer touching the codebase.
That intelligence surfaces in a drag-and-drop itinerary builder: ops teams pick charter and vendor products with live base, hourly, and daily rates, then assemble multi-day tours block by block — dropping travel legs, attractions, meals, and tour-leader slots onto a day-by-day calendar while the engine keeps pricing and availability in sync.
The latest leap builds directly on that foundation: a real-time AI voice agent sits on top of the same rules-and-inventory engine, so a traveler can simply describe the trip they want and watch a complete, priced itinerary come together live — no dragging, no forms.
Key Features
- AI Voice Itinerary Builder — the flagship: The platform's standout capability lets travelers plan an entire trip just by talking. A live AI voice agent holds a natural, back-and-forth conversation, understands what the traveler wants, and assembles a complete, priced itinerary on the fly — no forms, no manual scheduling, all in real time on the website.
- 4-Way Persona Dashboards: Each user group gets a tailored interface. Vendors manage their own inventory (charters, hotels), while students handle their own documentation and payments — all from role-specific views.
- Drag-and-Drop Itinerary Builder: A visual, calendar-based planner where ops teams build complete multi-day tours by dragging blocks — flights, charter legs, attractions, meals, and tour-leader slots — onto each day, with vendor pricing and availability wired in as they go. What was once a manual spreadsheet exercise becomes minutes of drag-and-drop.
- Automated Pricing & Margins: The system calculates business margins automatically based on real-time vendor availability and seasonal demand peaks, removing guesswork and manual markup.
- Financial Orchestration: Sophisticated payment routing collects funds from students and automatically triggers payouts to multiple vendors — while retaining Tango's profit margin — in a single, reconciled flow via Stripe.
The Impact
Delivered well within the academic cycle, the platform transformed Tango from a manual, spreadsheet-heavy operation into a high-tech marketplace — cutting manual operations by 50% and enabling seamless, automated payment routing. What began as a roadmap ambition is now live: AI-personalized, voice-built itineraries put trip planning directly in travelers' hands — and point to where the product goes next.
"The Rules Engine was the real intellectual challenge here — travel has hundreds of edge cases. We designed it as a composable rule graph so that Tango's ops team can add new pricing or refund logic without a single engineer touching the codebase."
Tech Stack
- Backend: Python (FastAPI) for high-performance APIs and orchestration
- Frontend: Next.js for a fast, mobile-friendly, multi-tenant UI
- Payments: Stripe for fund collection and automated multi-vendor payouts
- Core Logic: Custom automated Rules Engine built as a composable rule graph
- AI Voice Agent: Real-time conversational interface (speech-to-text, LLM-driven trip planning, and text-to-speech) that turns natural dialogue into a structured, priced itinerary live in the browser
- Architecture: Multi-tenant, 4-sided marketplace serving Students, University Heads, Vendors, and Internal Ops