The Industry View: How Travel Will Become AI-native
Summary
This panel session at ITB Berlin 2026, moderated by Dirk Rogl of Travel Commerce, brought together four senior technology and product leaders — James Byers (Product Lead, Travel, Google), Garry Wiseman (Chief Product & Technology Officer, Sabre), Piero Sierra (Chief AI Officer, Skyscanner), and James Waters (Chief Business Officer, Booking.com) — to debate how the travel industry becomes AI-native and what agentic commerce means for distribution, margins, and brand control.
The session opened with CEO quotes from each company's leadership, which were used to frame each panelist's strategic positioning. Sundar Pichai's quote that 'AI will fundamentally change how we find, understand and act on information — moving from links to agents that help get things done' set the tone for Google's role as an ecosystem connector. Sabre's CEO framed agentic AI as 'a new channel for booking,' and on the same day as the panel, Sabre announced both a company rebrand and the launch of Agentic APIs to support autonomous shopping, booking, and servicing workflows. Glenn Fogle of Booking Holdings described the current moment as 'the most radical reinvention travel has ever seen.'
A central tension the panel examined was the gap between AI enthusiasm and AI trust. James Waters cited data that today roughly 90% of travelers say they look forward to using generative AI, but when asked how confident they are to allow it to make decisions on their behalf, that figure 'falls into the single digits.' He argued this gap is structurally logical: travel is financially expensive, emotionally expensive, and time-expensive, meaning trust must precede delegation. He emphasized that trust is not just about the AI layer itself, but about the quality of supply information, payment security, and service reliability underneath it.
Piero Sierra, who transitioned from Chief Product Officer (a role he held for 10 years at Skyscanner) to Chief AI Officer roughly mid-2025 at the CEO's request, described a nuanced view of personalization: AI enables both per-user memory and per-trip context, but in-trip contextual personalization is currently dominating over individual user personalization. Skyscanner recently launched a ChatGPT integration that surfaces Skyscanner answers inside the ChatGPT interface, reflecting a strategy of being present on emerging AI surfaces while maintaining core product integrity.
James Byers discussed Google's Universal Commerce Protocol (UCP), announced at NRF in January 2026, as an emerging standard for agentic commerce — initially in retail, now being explored for travel. He acknowledged the travel and retail verticals have fundamentally different transaction dynamics, requiring careful adaptation. He offered a vivid illustration of the pace of change: 'If I said to you three months ago that the US would start to run out of Mac minis because everyone was running personal agents on them with access to their Gmail and passwords and credit cards, I think you might all look at me like I had maybe need a nap.'
Garry Wiseman described Sabre's AI-native ambition as holistic across plan, build, and operate cycles — including using Gemini for code generation and deploying AI bots for customer service. He also highlighted the risk of poorly constrained agents, citing a real example of a bot that purchased $400,000 worth of watches on behalf of a user, which had to be reversed.
James Waters disclosed that more than 60% of Booking.com's bookings business comes direct to its app or website, with additional large volumes acquired through Google. He noted that Booking.com recently launched a ChatGPT partnership and is working with PayPal on agentic payments — specifically highlighting installment payments as a key feature for expensive travel purchases. He predicted 2026 will see 'a more obvious, visual, dramatic evolution' of Booking.com's product than any other year in his 17-year tenure at the company.
On standards, Garry Wiseman called out the industry's need to engage in defining agentic commerce standards proactively, noting that Google is driving UCP and OpenAI has its own equivalent (ACP). He used the seat map as a concrete example of a travel-specific capability that is not yet represented in any current agent protocol. Piero Sierra added that with 'coding becoming free' — referencing a perceived capability shift in coding agents around November 2025 — stitching together across imperfect standards becomes more practical than it was under NDC.
Key Takeaways
- •Sabre announced on the day of the panel both a company rebrand and the launch of Agentic APIs, explicitly positioning itself as infrastructure for autonomous shopping, booking, and servicing — a direct response to agentic commerce becoming a new distribution channel.
- •James Waters cited a stark data point: approximately 90% of travelers express enthusiasm for using generative AI in travel, but single-digit percentages are confident allowing it to make decisions on their behalf — a trust gap all panelists agreed must be closed before autonomous agentic booking becomes mainstream.
- •James Byers (Google) predicted that fully autonomous AI agents acting on high-consideration travel purchases is 'probably an ITB 2027 topic more than 2026,' framing it as a structural timing constraint rather than a capability question.
- •James Waters revealed that more than 60% of Booking.com's business comes directly to its app or website, and committed that 2026 will see more visible product innovation at Booking.com than any of his 17 years at the company — with the new Booking.com x ChatGPT integration launching and a PayPal installment payments partnership going live at the end of the panel month.
- •Piero Sierra argued that in-trip contextual personalization (understanding the specific trip) currently dominates over per-user personalization (knowing the individual), and that AI enables both — but the trip context layer is delivering more value right now.
- •Garry Wiseman cited a concrete cautionary example: an AI agent autonomously purchased $400,000 worth of watches for a user incorrectly, requiring the transaction to be reversed — making the case that financial guardrails on agentic transactions are non-negotiable during the current learning phase.
- •Sabre is piloting Agentic APIs with both airlines and travel management companies in a human-in-the-loop mode where humans actively provide feedback to train and constrain agent behavior before expanding autonomous operation.
- •Piero Sierra transitioned from Chief Product Officer (10 years) to Chief AI Officer at Skyscanner around mid-2025 at the CEO's explicit request, because AI was 'affecting every layer of the company' — products, internal productivity, engineering practices, and industry partnerships.
Transcript
Yeah, and let's paint the picture of our industry now. If I agents increasingly plan, compare, and execute travel decisions as we learned, what does it really mean for our industry to become AI native? It's about chat bots now and it's about future. It's about interfaces, but it's also about control, about relevance, and about trust. And that's very human words, isn't it? So who owns the traveler relationship when decision when decisions are delegated to machines and what happens to margins to d...
