More than Automation: Localization Becomes a Rich Experience with BWX
Gabriel Fairman, founder and CEO of Bureau Works, is a “young veteran” of the language services industry, having set up a translation business in 2005 that became a technology company in 2021. During a SlatorCon Remote presentation on November 29, 2023, hosted by Slator’s Andrew Smart, Fairman shared with attendees his unique perspective on the role of technology in every aspect of the translation workflow.
To Smart’s question about some of the challenges that companies face in addressing and incorporating AI into their businesses, Fairman replied that to him, the number one challenge is actually communication around what AI is, what it isn’t, what it can do, what it can’t do. He added that he stays away from the AI acronym because it can mean very little and a lot at the same time, as well as many different things for different people.
Fairman focused instead on what can be accomplished with AI, such as context-sensitive translation, “something that the large language model can handle,” said the CEO. He added that when people get past the idea of what AI is or isn’t and start looking at it from a more practical perspective, a lot can be done by it, which is what he’s witnessed with BWX, as Bureau Works tool is called, including translating 50% faster.
The question he asked himself about the platform was, “what can we make for a better, more augmented experience for the translator?” And the answer that emerged was that, yes, “I think we can already do a lot with it and it’s a lot of fun.”
BWX, All-in-One with AI at the Ready
Fairman took attendees through a tour of BWX, describing the platform first as “an alphabet soup … a TBMS, TMS, all in one … the management environment where you can do everything from managing your projects to paying your people to keeping track of quality.” He then showed how the AI-assisted translation functionality can be enabled or disabled as necessary, adding that contents are private to the user in BWX, and that the app can also be pointed to a private instance of OpenAI or Azure.
The editor window in BWX, which Fairman shared after explaining the many project management options available in the platform, shows translation in action. Translation memory matches and machine translation segments are displayed in different colors in an attractive user interface, with terminology suggestions popping up as the linguist edits. There is also a preview window to show progress in context.
Another useful functionality in BWX is a QA step that happens as a segment is finalized, which in the app is called “translation smells” and points out any linguistic issues that might need correction. Fairman described this as a dialogue between the linguist and the platform, with no prompting involved, no need to have an awareness of the language model, which does its job in a way that makes sense. Additionally, the engine is learning as the linguist works.
Fairman pointed out that the AI functionality in BWX can actually highlight not just potentially incorrect translations, but also issues of style like wordiness or of culture, such as inappropriate translations, along with gender bias and a lot more that a translator might not be aware of. The AI can also make suggestions for alternative translations.
Customizing Every Experience
Fairman answered a number of very good questions posted by curious attendees, addressing for example the flexibility to adapt BWX to the needs and wants of different people at work. Fairman acknowledged that “everybody has their own different methodology, their preferences,” and BWX can accommodate them all.
Fairman summed up the way translation work goes for linguists inside BWX as a partnership between the human and the platform that not only increases productivity, but also provides for “a more pleasant experience translating.” It is the opposite of the post-editing task dreaded by many and instead offers the linguist a rich experience.
Automation happens both as a result of the capabilities embedded in the platform and the many possible ways in which it can be customized to fit scalability and efficiency needs at the enterprise level, from workflows to the very interaction between TM, MT and AI, to connecting various applications via API, explained Fairman. He added that “you can automate your quotes, you can automate your billing, you can automate your payables, you can automate quality management, and, you can automate your job placements.”
To book a BWX demo, visit the Bureau Works site.