Fairman challenged that traditional view of language, where it is mere representation, instead emphasizing its role in identity formation, as a factor driving outcomes and as a tool for action. “Language isn’t about just representing things. Language is about getting people to do certain things like buying, sharing, liking, commenting, understanding,” he said.
Linking Language to Outcomes
To Fairman, a shift toward a less “transactionalized” view of language necessitates a new understanding of quality, moving beyond linguistic accuracy to focus on efficiency in achieving desired outcomes. This includes factors like engagement, conversion rates, and time spent on page.
He sees a fundamental division in how technology is developing. On one side, there’s a push for “black box” solutions that prioritize precise outputs without concern for the process, and on the other, there is a more human-centric approach where technology aids and collaborates with humans.
Then, within the human-centric approach, there is further distinction between passively involving humans in the process and actively empowering them to own the outcomes. Fairman advocates for the latter as a transformative approach.
In that context, Fairman introduced the concept of agentic translation, differentiating it from AI assistants. While assistants react to user prompts, agents operate proactively in the background, providing insights and recommendations.
He showcased examples of Bureau Works’ agents, such as the terminology learning agent and the “Smells” agent, which identifies potential issues like inconsistency and bias. He also highlighted the “Sous Chef” agent, which enables the creation of multiple content variants tailored to specific personas.
“It’s complicated for us to imagine, like a single website running, for instance, with hundreds of different variants. Now we’re talking about a world where that’s possible, where we’re talking about content that’s very finely tuned to the way in which a reader approaches that text.” — Gabriel Fairman, CEO at Bureau Works
Expanding on the “Smells” agent, Fairman said that it runs in the background, “looking at a very wide range of potential issues with the translation. It’s flagging those to the translator. This is an example of a semantic verification agent. But the opportunities for agents are quite limitless.”
Fairman emphasized that these agents empower human authors to make informed decisions, moving beyond the limitations of traditional translation workflows. He envisions a future where authors, equipped with AI-powered insights, can optimize content for specific audiences and achieve measurable results.
“It’s complicated for us to imagine, like a single website running, for instance, with hundreds of different variants. Now we’re talking about a world where that’s possible, where we’re talking about content that’s very finely tuned to the way in which a reader approaches that text,” added the CEO.
To Fairman, such a level of optimization just was not attainable with the previous technology, and now there is an opportunity to link textual decisions to outcomes.
Fairman concluded by stressing that the role of the translator is evolving into that of an author, responsible for crafting content that resonates with specific audiences and drives desired outcomes. This shift requires new tools and a renewed focus on creativity and strategic thinking.
