Travel Chat Agents in 2026: Expedia Romie, Trust Gaps, and the 8% Booking-Authority Problem
By Sagar Shankaran, Founder of CallSphere
Expedia's Romie reshuffles itineraries when flights delay, but only 8% of 2026 travelers grant AI booking authority. Here is the build pattern travelers actually use.
Key takeaways
Expedia's Romie reshuffles itineraries when flights delay, but only 8% of 2026 travelers grant AI booking authority. Here is the build pattern travelers actually use.
What broke in travel chat before AI?
flowchart TD
WA[WhatsApp] --> Hub[Channel Hub]
SMS[SMS] --> Hub
Web[Web Chat] --> Hub
Hub --> Router{Intent}
Router -->|book| Booking[Booking Agent]
Router -->|support| Support[Support Agent]
Router -->|sales| Sales[Sales Agent]
Booking --> DB[(Postgres)]
Support --> KB[(ChromaDB RAG)]
Sales --> CRM[(CRM)]Travel support has been a phone-and-email channel for two decades. A flight gets delayed, the OTA emails a 14-line update, and the traveler is on the phone with the airline at the gate trying to get rebooked manually while the OTA's "manage trip" page shows stale data. Hotel front desks tried web-chat widgets that hibernated overnight; airlines built mobile-app chat that was actually a forwarder to the same hold queue. The traveler was the integration layer.
Itinerary apps fared better but stayed read-only. TripIt aggregated; Google Flights tracked. Neither could rebook a leg, find an alternate hotel near the new arrival airport, or reshuffle a dinner reservation when the flight pushed to tomorrow. The conversational gap — describe what happened in plain English, get a reshuffled plan back — was what travelers wanted and no product fully delivered until 2026.
What chat AI changes (specifically)
Expedia's Romie, announced as an AI virtual concierge, monitors weather and last-minute disruptions, aggregates travel info from emails, suggests activities and dining options near the hotel, and updates the itinerary in real time. Expedia's ChatGPT-app integration lets travelers plan trips inside ChatGPT itself, with cross-channel state. Travel Weekly's coverage of Expedia and Booking putting AI through its paces shows the category is live, not pilot.
Hear it before you finish reading
Talk to a live CallSphere AI voice agent for hospitality in your browser — 60 seconds, no signup.
The honest 2026 number is the trust gap: a NomadLawyer-cited survey found only 8% of US travelers willing to grant AI booking authority, while 43% want AI to help compare options and 37% want AI alerts on price drops or schedule changes. Translation: travelers want copilots, not autopilots. The right design ships AI as the comparison and alert layer, with a one-tap "yes, book it" affirmation that keeps the human in the loop.
How CallSphere applies this
Explore a live demo and compare current plans to find the right fit for your business.
Build/launch steps
- Define the v1 scope as alert + reshuffle, not autonomous booking. Travelers buy this; they reject autopilot.
- Wire the PNR or itinerary API so the agent reads current state in real time.
- Build the disruption trigger — flight delay, cancellation, weather, hotel overbook — and define the message template per case.
- Make the rebook step explicit and one-tap; the agent presents options, the traveler taps to confirm.
- Wire SMS and WhatsApp as the primary push channels; chat-on-site is the secondary surface.
- Drop the agent from /embed and ship a pilot with one route or one hotel.
- Track disruption-recovery time, traveler CSAT during disruptions, and rebook-success rate weekly.
ROI / KPI section
The KPI travel operators care about is disruption-NPS — net promoter score from travelers who actually experienced a delay or cancellation. That is the moment churn gets decided. Mature deployments lift disruption-NPS by 15–30 points by replacing the 30-minute hold queue with a 30-second chat reshuffle. Secondary KPI: rebook-success rate (travelers actually rebooked through the agent rather than calling the carrier).
FAQ
Q: Should the agent book autonomously? A: No. Per the 2026 NomadLawyer survey, only 8% of travelers grant booking authority. Ship the comparison and alert; let the human tap to confirm.
Still reading? Stop comparing — try CallSphere live.
See the hospitality AI agent handle a real call — complete, industry-specific, and live in your browser. No signup.
Q: Does CallSphere support multilingual disruption flows? A: Yes — 57+ languages on the platform. Critical for international travel where the disruption hits during the wrong time zone for the traveler's home language CSR.
Q: What about ChatGPT travel apps? A: They are complementary. Travelers will plan trips inside ChatGPT, then use your branded chat for in-trip support. Build for cross-channel state.
Q: Can the agent rebook with a held PNR? A: Yes, with explicit confirmation. CallSphere's tool layer supports write actions with an audit trail.
Explore a live demo and compare current plans to find the right fit for your business.
Sources
Written by
Sagar Shankaran· Founder, CallSphere
LinkedInSagar Shankaran is the founder of CallSphere, where he builds production AI voice and chat agents deployed across healthcare, hospitality, real estate, and home services. He writes about agentic AI, LLM engineering, and shipping voice agents that handle real calls in production.
Try CallSphere AI Voice Agents
See how AI voice agents work for your industry. Live demo available -- no signup required.
