“Scaling AI Requires the Courage to Dismantle”: Lessons from TUI

As the Technology Team Lead at TUI Group, Jie Zheng operates in the layer most AI conversations skim past: execution. Her remit cuts across systems, teams, and processes shaped by decades of operational build-up, where AI efforts either compound or stall out. 

Ahead of her session at Skift Data + AI Summit 2026, we asked her how AI is actually landing inside operations, what leaders still get wrong, and how to measure value when the playbook is still being written.

1. What has surprised you most about applying AI at scale in your organization?

“The most striking revelation has been the sheer complexity of Business Process Re-imagination. While it is widely understood that AI cannot simply be layered onto existing workflows, the reality is that most operational processes have matured organically over decades. 

It is a classic architectural dilemma: do you build a pristine road system designed in a lab, or do you pave over the ‘desire paths’—the footpaths well-walked by teams over time? The truth sits somewhere in between. 

Scaling AI requires the courage to dismantle these entrenched, organic architectures while remaining pragmatic enough to deliver incremental value. Ultimately, the challenge is less about the code and more about a fundamental cultural transformation—one that balances visionary creativity with the relentless rigor of disciplined delivery.”

2. Are there common misconceptions about AI in travel operations that you would like to clarify?

“A prevailing myth is that AI will immediately enable hands-off autonomous operations. In reality, scaling AI in travel requires more human oversight, not less—at least initially. 

We view Co-Pilot capabilities as significantly more mature and implementable than a full autopilot model. While technical demos of autonomous systems are often impressive, real-world implementation at scale remains largely theoretical, particularly concerning their long-term impact on business resilience and performance. 

Ultimately, AI should empower our teams to filter out the operational noise and refocus their expertise on the high-value exceptions that truly define the guest experience.”

3. How do you approach measuring AI’s value without giving away all your company strategies?

“Rather than framing AI as a standard central IT project with a standard ROI, it is more effective to consider how different business units justify their budget and spending during board meetings using their own native metrics.

For instance:

• A Chief Product Officer might justify investment through direct revenue growth, cost efficiencies, and a significantly accelerated time-to-market for new features.
• A Chief People Officer might focus on ‘time-to-competence’ and workforce resilience in high-attrition roles.
• A Chief Innovation Officer might point to the option value of new capabilities and the velocity of experimentation.
• A Chief Information Security Officer evaluates investment through the lens of risk surface reduction and automated threat response.

I would encourage adopting a multi-dimensional framework tailored to the company’s specific maturity stage, allowing for a comprehensive view of value that evolves with the technology.”

4. What is one piece of advice you would give travel leaders about balancing AI and human judgment?

“My advice is to learn from the corporate innovation framework and walk on two legs: allocate a dedicated R&D budget for ‘true creativity’—experimental ventures where you can afford to fail in pursuit of breakthroughs—while directing the majority of your resources towards tangible, implementable deliveries, with AI serving as an augmentation rather than a complete replacement. 

In such a volatile landscape, being a disciplined ‘fast follower’ is often the most resilient and balanced approach. 

Once a use case or capability is proven—not just technically, but also in terms of the team and the ‘ways of working’ required to sustain the solution—you can then move decisively to integrate and scale it, whether through internal builds, third-party acquisitions, or a hybrid model.”

Continue the Conversation at Skift Data + AI Summit 2026

Operators scaling AI inside real-world travel systems will recognize both the tension in these ideas and the opportunity within them. Jie Zheng will unpack what this looks like in practice at Skift Data + AI Summit 2026.

Join the industry’s leading data and AI decision-makers this June in New York City.