General-purpose personal AI agents are beginning to book and manage trips from the full context of a person’s life. That shift changes where travel demand and customer relationships live — and leaves traditional travel companies mainly handling fulfillment.
Personal AI agents (ChatGPT, Claude, Gemini, Instinct, Town, Meta’s Muse) are becoming capable of booking, rebooking, and managing travel by using a person’s calendar, email, loyalty accounts, and preferences.
This agent-first model captures the demand-signal conversation and customer authority, potentially shifting travel companies into a fulfillment role focused on plumbing and execution.
Widespread adoption is still early: tools are brittle, expensive to automate at scale, and raise unresolved security, privacy, and accountability issues when mistakes create costly outcomes.
# The basic shift
# Why that matters for travel companies If a personal agent owns the conversational relationship — the planning, tradeoffs, and booking authorization — travel businesses risk being relegated to the transaction and the fulfillment plumbing (reservation systems, ticketing, customer service). The valuable demand signal — the nuance of why and when someone wants to travel — could live inside the agent, not with the airline, OTA, or hotel.
# How we got here Previous attempts at personal concierges failed because the interface was right but the economics and technology were wrong. Now large, general-purpose models and startups with deep inbox and browser integration can search, compare, and complete bookings. Startups like Instinct and Town moved fast by connecting directly into email, calendars, messages, and browsers. Major models (ChatGPT, Claude, Gemini) are heading the same way.
# What's still uncertain Adoption remains limited to power users. Browser automation remains brittle and costly. More fundamental issues remain unresolved:
These gaps slow mass adoption and create operational risk for travel brands that embrace or block agents.
# Immediate implications for travel businesses Treat the agent as a potential front door to customers. Companies should decide how much of the guest relationship they expect to run through third-party agents and whether internal organization (marketing vs. service siloes) will hinder that integration. If agents capture demand signals, travel companies need strategies to preserve access to customer intent, either by embedding into agents or by providing clear, agent-friendly interfaces and APIs.
# Practical steps to consider now
# Bottom line
Expedia wants a seat inside the new wave of AI agents, including Meta's Muse. But CEO Ariane Gorin is betting on a messier future for travel: multiple entry points and specialized agents, not one bot that owns the whole trip.
Travel brands have to evaluate if they're structured to let an agent handle the full relationship, and whether internal divisions may hold them back.
Almost every travel brand CEO at Skift Global Forum touched on what they want AI agents to do. Fewer talked about how they expect it to work.
Instinct, a freshman AI assistant that has Silicon Valley buzzing, can book and rebook trips from a text thread. Key lesson for travel: Talking about frictionless booking isn't the same as delivering it.
Stuart Gentle Publisher at Onrec 23 Sep 2026 | How Technology Is Changing the Way We Travel Technology has totally transformed the way we travel. What once required phone calls, paper maps, travel agents, and printed tickets can now be handled from a smartphone. Travelers today can compare prices instantly, book flight
Skift Global Forum spent three days talking about AI agents, digital concierges, and connected trips. All of it assumes the traveler is always online.
Loading more related stories...
Open the app view to save this story, compare related coverage, and continue from the same source.