Intelligence Track

Design

2 July 2026

The Brief

Trust' became online travel's favorite word this week — and two camps meant opposite things by it. The incumbent giants are betting travelers keep returning to known brands because a general-purpose model might hallucinate a booking; a larger rival is betting the real contest is earning the trust of the AI agents themselves, exposing verified inventory and live pricing as machine-readable rails so demand routes through it whether or not anyone opens an app. That split is the actual story, and it rhymes with everything else on the wire: a payment network moving upstream from the swipe into curated discovery, a global leisure platform walking into managed corporate travel just as the domestic incumbent that owned that lane slides toward a delisting notice, a state rail monopoly absorbing premium-demand growth through a channel it controls and is mid-rebuild on. Even a booking assistant's personality — how it hedges, whether it asks before acting — turns out to be an unowned trust surface no one deliberately drew. The question for the category: when discovery, settlement, corporate supply, and the agent's own voice each belong to someone else, what is the aggregator still the owner of?

UX Collective · 2 Jul 2026

Assistants built on the same model and instructions still feel like different people — one hedges and defers, another acts then reports — and the argument is that these traits emerge as an accidental byproduct of alignment, not deliberate design. The proposal: treat AI personality as an interface layer chosen the way a team chooses a button state, rather than letting it arrive undrawn.

Reading as

No items for this audience filter.