News

31.08.2023

Lucas Seiler

More Efficient Translation: How Specialised AI Translation Saves Money

Your language, your style, your AI translation. With specialised AI translation, you get precise translations tailored to your requirements and preferences. The result? The highest translation quality, minimal need for corrections and significant cost savings. Learn more about how Textshuttle teaches artificial intelligence your corporate language.


Reduce costs with specialised AI translation

Specialised AI translation goes far beyond the mere correct translation of texts. It adapts to the company’s specific requirements by incorporating the finest of nuances and precisely coordinating the tone, phrases and terminology to match the company’s corporate language.
The translation quality of specialised AI translation is so high that the translations generated by it require barely any correction at all. This has been shown by regular studies, not only by us, but also by our customers: one multi-year study by a customer, for example, found an efficiency gain of 50% after just two years.
For larger Swiss companies that have to communicate in all of Switzerland’s national languages and English at a minimum, this can amount to annual savings up to the seven-figure range. Even with lower translation volumes, however, cost reductions can quickly add up to several hundred thousand Swiss francs. And because AI translation continuously learns and improves, the savings potential continues to rise over time.


Our approach compared to generic translation

In the development of specialised AI translation, incorporating a company’s existing language data, and in particular previous translations, in the development process is absolutely critical. Why is this so important and how does the Textshuttle approach work?
All common, state-of-the-art AI systems are based on learning from a multitude of examples, or what is known as training data. In the case of AI-aided translation, the training data is texts and their translations. The learning process enables the AI to translate new texts based on the training data used during the training process. This training data is crucial for the translation quality.
Generic translation systems use data from many different sources. For the end user, the origin of this data is not transparent; they can neither control the quality and selection of the training data nor specify a preference for certain text types. The result is translations that reflect a cross-section of all of the sources – a generic-sounding language. Although generic translation solutions can certainly generate solid translations, they fail to take into account companies’ exacting requirements – such as the incorporation of their specific corporate language.
The Textshuttle approach, by contrast, places your company’s extensive language knowledge, painstakingly built up over years of work, at the centre of the AI: through the systematic integration of your previous language data in the training process, the AI system is specifically taught your corporate language – in short, the system learns your language. Your style guides and terminology are also included as additional sources in the translation process.
The result is translations that are noticeably shaped by your corporate language and meet the highest quality standards. This quality difference is by no means merely theoretical – the practical results are evident. This has been shown by numerous blind tests by our customers in comparisons with other commercially available translation systems.

Chart_Welche Übersetzung is besser_EN.png

Qualitative blind comparison by Textshuttle customer from the insurance sector (December 2022)


Development in close collaboration

In the development of specialised AI translation, we work closely with your company’s language professionals. At the start of the project, we meticulously analyse your company’s specific requirements and wishes. Then we train an initial specialised AI system using your language data. Your company’s experts are actively involved in the evaluation of the translation quality and can give direct feedback and suggest changes along the way. The result is an AI system that translates precisely according to the preferences and specifications of your company.


Continuous improvement

Language is not a static thing, but in continual development. That’s why we place great importance on updating your AI translation with the language data you’ve generated in the meantime at regular intervals after the initial specialisation. This ensures that the AI system is always up-to-date. These updates of the AI translation also make it possible to learn from previous translation errors and continuously improve. So your AI translation will get even better over time.


More information

Would you like to learn more about the possibilities of specialised AI translation in your company, get a no-obligation assessment of your business case, or have us evaluate the savings potential for your company with AI translation? Then contact us through the contact form. We’ll be in touch.