An AI concierge for the worst day of a guest's trip.
A working prototype showing how Virgin Australia could use an LLM-powered assistant to handle flight disruption end-to-end — without putting guests in a queue.
The problem
Disruption is the moment guests judge an airline.
When a flight is cancelled or delayed, the existing experience funnels guests into long queues, hold music and self-service flows that weren't designed for the emotional weight of "I'm not going to make it home tonight." Most of the work — looking up the booking, finding the next service, organising a hotel, sending a confirmation — is repetitive and well-suited to automation.
The solution
A single chat that does the work for the guest.
An LLM agent fronts the experience and a set of typed tools do the real work — looking up the passenger, finding alternative flights, issuing hotel vouchers, and composing the confirmation message. The guest never sees the plumbing; they just see a calm, Australian voice that takes ownership and resolves the situation.
Conversational front door
Guests describe what's happened in plain English — no forms, no IVR.
Agent + tools
The model orchestrates booking lookup, rebooking, accommodation and messaging tools.
Always on tone
Tone, policy and escalation rules are baked into the system prompt, not left to chance.
Minutes, not hours
End-to-end recovery in a single conversation, while staff focus on the cases that need a human.
How it works
One flight number in. A sorted guest out.
Guest shares a flight number
The only thing the assistant asks for. Nothing more, nothing less.
Agent diagnoses & rebooks
It names a specific cause, apologises, calls the rebooking tool and presents 2–3 clear options.
Plan confirmed in writing
Hotel and meals are arranged if needed, and a written confirmation is sent — no follow-up call required.
Want to see it in action?
Jump into the live prototype and try a disruption — any flight number will do.
Let's see it