.svg){kind=icon}
About the Client
We worked with a key travel technology company.
The Challenge
They were working with a major airline who needed around 25k location names (such as cities, towns, airports etc.) translated into 6 different languages in a short space of time, to benefit their customers when travelling between different regions.
How We Helped
We built them a tool along with user guide and documentation, which uses a trusted data source combined with AI to process all the translations.
It reads in their existing data sets, transforms that into something efficient to work with to match with the vast data set (over 10 million location names) to find all available translations, and then uses AI to translate them all again and take the best output by combining both, running quality control checks at each stage before transforming this into files compatible to upload directly into their system.
What We Achieved Together
This approach proved to be the most comprehensive way to translate location names while maintaining a level of trust, because we don’t rely entirely on AI (which, as we all know, “can make mistakes”). It gives us data to quality check against the AI and make decisions about the result, for example preferring the AI output over the data set where the two differ.
What This Means Going Forward
The best approach to getting trustworthy results is not to blindly trust AI, but to use it precisely where it can add benefit, often augmented alongside other methods to provide the best solution.
