Chief AI Officer or CTO? How Travel Leaders Should Organize for AI
AI is now table stakes in travel. According to Phocuswright, 83% of companies are using generative AI, 13% are exploring use cases, and just 1% have no plans at all. That level of adoption forces a structural decision: Do you appoint a chief AI officer, or keep AI inside the CTO's remit?
We're seeing both moves. Expedia Group named Xavier Amatriain its first chief AI and data officer. Airbnb appointed Ahmad Al-Dahle as CTO, emphasizing his AI depth. The split isn't about hype-it's about how you build ownership, ship real value, and manage risk.
The strategic fork: AI as transformation vs. AI as core tool
American Express Global Business Travel (Amex GBT) treats AI as a company-wide capability with clear owners. David Thompson, its chief information and technology officer, and Evan Konwiser, the chief product and strategy officer, share oversight. They run AI like a product line: measurable objectives, a budget, VP-level ownership, and a roadmap.
Kayak keeps AI inside the CTO. Yaron Zeidman's stance: a CTO-led model scales AI across products and operations, as long as the work is integrated, not siloed. AI isn't a side project-it's embedded in product, data, and infra decisions.
Skyscanner created a CAIO role. Piero Sierra owns AI vision and strategy, tracks the tech pace, codes alongside teams, and coordinates execution across the company. Direction is centralized; delivery is distributed.
When a CAIO makes sense
- Your AI ambition requires visible transformation across product, data, platforms, and partnerships.
- Leadership needs a single owner to set direction, align roadmaps, and prevent "AI as everyone's side task."
- The tech is moving faster than your current operating cadence; you need someone full-time on horizon scanning, standards, and sequencing.
- You sell to enterprises that expect a named AI executive for credibility and governance.
When the CTO should own AI
- Your value is delivered through product and platform, and AI is already part of engineering, data, and MLOps.
- You can enforce integration across teams without standing up a separate org that risks silos.
- Your leadership bench (VPs/Directors) can carry AI objectives with clear OKRs and budget.
- You want continuity: one technology leader accountable for build vs. buy, model choices, and scale.
What boards should demand-regardless of structure
- Single point of accountability: Name who owns AI outcomes and risk. Everyone contributes; one leader is answerable.
- Investment thesis with ROI: Track internal productivity wins and revenue impact by use case, not by tech fascination.
- Risk and trust framework: Data governance, model provenance, evaluation, security, red-teaming, and human-in-the-loop.
- Sequencing: A 12-month roadmap that moves from obvious wins to durable platform bets.
- Metrics: Adoption, latency, quality scores, cost per inference, and business outcomes tied to P&L.
The Amex GBT approach: structured innovation with accountability
Thompson's remit spans technology and risk, internal ROI, and sequencing. Under Thompson and Konwiser, AI work sits with named leaders at the VP level, with measurable goals and budget. As Marilyn Markham, VP of enterprise AI strategy, puts it: build capacity to adapt; don't wait for the tech to "settle."
The Kayak stance: embed AI, avoid silos
Zeidman treats AI as part of the core toolset-alongside UI, infra, and data. The test: If AI isn't in the product and engineering bloodstream, it won't scale. A separate AI strategy that lives outside delivery teams slows you down.
The Skyscanner model: a CAIO to set direction, with distributed execution
Sierra's focus is clarity and speed. He owns vision and prioritization across product, partnerships, and platforms, keeps data and tooling aligned, and works with teams hands-on. The role is reassessed annually-because needs will change as the tech shifts.
How AI is changing the rest of the C-suite
- Chief Product Officer: Translate AI capabilities into features customers will pay for; prioritize by value, not novelty.
- Chief Legal Officer: Keep pace with evolving regulations, data residency, copyright, and model licensing.
- COO: Redesign processes for AI-native workflows; measure cycle-time reduction and quality lift.
- All executives: Use AI as a thinking partner for analysis, drafts, and decision framing. Literacy is now leadership hygiene.
A 90-day plan to move AI from side project to system
- Days 0-30: Inventory current use, shadow IT, and data flows. Ship baseline guardrails (privacy, safety, IP). Stand up evaluation metrics and a lightweight model registry.
- Days 31-60: Prioritize 5-7 use cases by impact and feasibility. Pick platform enablers (vector store, evaluation, observability). Define decision rights across product, data, security, and legal.
- Days 61-90: Launch 2-3 pilots with owners, OKRs, and budget. Instrument everything. Set a quarterly operating rhythm for review, scale, or kill.
Operating model that actually ships
- AI Council (small): CITO/CTO or CAIO, Product, Data, Security, Legal. Meets biweekly. Decides standards and sequencing.
- Delivery pods: Product + Eng + Data + Design + Domain SME. Accountable for outcomes, not demos.
- Risk by design: Threat modeling, eval suites, bias checks, and human oversight in the workflow-not in a separate review gate.
- Vendor sanity: Clear integration criteria, exit plans, and total cost tracking (infra + tokens + people time).
How to choose-three blunt questions
- Where will AI create the most value in the next 12 months: customer experience, operations, or new revenue?
- What are the top risks you must control: data leakage, compliance, IP, brand safety, or reliability?
- Which structure gives you speed with accountability: a CAIO setting direction across functions, or a CTO embedding AI everywhere?
What's next
Some leaders expect AI-specific roles to fade as literacy becomes default. Others see clear need for a named executive during this transition. The common thread: focus and ownership now, with an annual review of structure as your capabilities mature.
The end state is simple: AI disappears into how work gets done. Not because it's gone-but because it's everywhere, and everyone knows how to use it.
Further resources
Your membership also unlocks:
