What does it take to make AI feel natural across languages, cultures, and markets?
In this edition of Behind the Mic, we speak with Lucía Guerrero, AI Program Manager, Globalization at Expedia Group, about the evolving role of AI in global customer support. From real-time translation and cultural adaptation to the importance of localization in AI development, she shares how technology and language expertise can come together to create more meaningful customer experiences worldwide.
Q.1 What motivated you to pursue a career at the intersection of AI, globalization, and customer experience?
My path began with a background in translation and computational linguistics, along with a curiosity about how technology can help people communicate across languages. That led me to machine translation in the early 2010s, where I worked with rule-based, statistical and neural approaches and built and led programs spanning text and transcription. When generative AI emerged, I quickly incorporated it into my existing programs and soon moved into customer support. I was drawn to the opportunity to bring together the parts of AI program leadership I find most energizing: experimentation and expectation management. Customer support is real-time, so it requires continuous learning and adaptation as the technology evolves. At the same time, creating a strong customer experience means balancing quality with latency and guardrails, while aligning stakeholders and linguistic teams on what quality means and where trade-offs are acceptable. That combination of language technology, practical innovation and meaningful customer impact is what motivates me.
Q.2 What are some of the most significant challenges involved in expanding AI-powered customer support across languages and markets?
Three challenges stand out. I heard a line repeated at several CACES Munich sessions: “Agentic AI is easy to demo and hard to deliver”. The technology is advancing quickly, and I’m particularly impressed by the state of the art in voice. But AI is nondeterministic, so pre-launch testing can’t cover every possible scenario. We need to identify what matters most, monitor performance continuously and build feedback loops that help us respond as real-world issues emerge. A second challenge is cultural fit. Many models are shaped primarily by English-language data and, often, US cultural norms. A tone that feels engaging to a US customer may feel unprofessional to a German customer, because German audiences often prefer direct and less enthusiastic messaging. Translators and localization experts can work closely with machine learning teams to evaluate and adapt model outputs so they feel right for each market. Finally, when similar technology is widely available, customers may distinguish companies by how the experience makes them feel. Translators can help make interactions meaningful and natural in each market, rather than impersonal or generic.
Q.3 Could you share a success story from your work in AI and globalization, highlighting its impact on customer experience or support operations?
A specific example is real-time chat translation for Japanese and Korean, two of our focus-market languages. In these languages, a good customer experience depends not only on accurate translation, but also on using the right level of formality and honorifics, choosing an appropriate degree of directness, and balancing warmth with professional distance. Neural machine translation (NMT) has been widely used to help agents communicate with travelers and partners across languages. Although it has improved significantly, it can still fall short on these nuances, even when a model offers formality settings. That can lead to an uneven experience and require frequent manual adjustments to glossaries and rules. Moving to large language models (LLMs) for real-time machine translation has been a game changer. By combining prompts with retrieval of our style guidelines, we can better guide tone and formality. We’ve seen an increase in linguistic quality scores, and we’re now assessing how that translates into the customer experience.
Q.4 How do you collaborate with localization, engineering, product, and customer support teams to ensure AI solutions meet the needs of customers across different languages?
As a Program Manager, I lead initiatives that are deeply cross-functional; not only between Localization and Engineering, Product and Customer Support, but also across our own Localization team. At Expedia, we favor working in squads, which means bringing the right partners together around a shared customer problem rather than relying on a series of handoffs. My role is to create alignment on the use case, the needs of each market and what success looks like from a technical point of view. Product and Engineering help shape a solution that is feasible and fits the customer journey, Customer Support brings insight into customer pain points and Localization specialists help ensure the conversation is accurate, natural and appropriate in each language. I see Localization as a partner throughout the process, not just a final review step. We bring language and market expertise into decisions early, evaluate the quality of AI-generated experiences and use our continuous linguistic quality assurance, as well as feedback from customers and support teams, to identify where the solution needs to improve. That ongoing collaboration helps us balance technical feasibility with a locally relevant customer experience.
Conclusion
Lucía’s insights highlight how successful AI adoption in global customer support goes beyond technological advancements. It requires understanding cultural nuances, maintaining linguistic quality, and bringing localization expertise into the process from the very beginning. From evolving machine translation with LLMs to building stronger cross-functional collaboration, her experience demonstrates how AI can create more natural and meaningful customer interactions across markets.
As AI continues to reshape customer experience, Lucía reminds us that technology may power the conversation, but understanding the customer makes it meaningful.
"Explore more perspectives from industry leaders at the Conversational AI & Customer Experience Summit (CACES), where experts share ideas and experiences shaping the future of AI and customer interactions. Stay tuned for more Behind the Mic interviews and industry insights on our blog!"
