This Singapore founder is building an AI translation company that knows when not to use AI

Daphne Tay, founder of Bluente, a Singapore-based company specialising in AI-powered document translation, has made a rather curious decision for someone whose business operates at the frontier of artificial intelligence. “We decided to keep the document-processing part of Bluente non-AI,” she explains, acknowledging that “for a company building an AI-powered translation product, that might sound counterintuitive”.

This reasoning reveals a great deal about how the young entrepreneur approaches an industry increasingly enamoured with its own possibilities. For Tay, the distinction between translating language and preserving a document’s structure is fundamental. While AI excels at interpreting meaning, elements like tables, columns, and clauses demand absolute predictability.

Consequently, she opted for traditional engineering to handle those structural components, resisting the temptation to introduce AI simply because it was available. It is a remarkably grounded philosophy for a founder whose generation came of age amid a technological revolution. 

Then again, Tay has never seemed particularly interested in innovation for its own sake.

Something lost in translation

The idea behind Bluente emerged during her time at Bain, where she discovered how little conversational fluency prepared someone for business’ specialised vocabulary. Struggling to familiarise herself with Chinese terminology, she began contemplating a solution that would eventually become her company.

Initially, the business concentrated on language learning. But as Tay encountered customers operating across different markets, she discovered that linguistic proficiency was only one part of a more complicated professional frustration.

“Many customers had a more immediate need,” she recalls. “They needed to get their work done across languages, sometimes under very tight deadlines.”

A document might emerge from translation with its words intact, yet leave the recipient confronting disorganised tables, misplaced figures, or formatting that demanded hours of reconstruction. The technology had ostensibly done its job while leaving much of the real work unfinished.

For Tay, this exposed a distinction that the enthusiasm surrounding AI sometimes obscures. “I reject the assumption that once AI can translate the text, the problem is largely solved,” she insists. The structure of a document carries meaning, too, particularly when a misplaced number or altered clause could change how somebody interprets its contents.

More consequentially, she argues that organisations must distinguish “between automating a task and delegating judgement.” Technology might translate a contract, for instance, but the professional reviewing it remains responsible for understanding its implications.

The virtue of restraint

For a young founder navigating an industry whose technical capabilities seem to expand almost daily, commercial pressure to keep adding more is considerable. Customers want increasingly comprehensive solutions, while competitors continually introduce new features that promise to transform familiar workflows.

Tay, however, has become wary of equating expansion with improvement.

“There’s a commercial incentive to add capabilities that grow contract value,” she acknowledges, “but every addition also creates a new promise we have to deliver on.”

The observation reaches beyond software development into the harder business of managing expectations. Once an organisation integrates technology into its daily operations, reliability becomes a commercial obligation rather than a desirable technical characteristic.

Sometimes, the problems worth solving are almost embarrassingly ordinary. “A formatting problem might look minor to us but create hours of work for someone else,” Tay observes, and she wants her team to retain an intimate understanding of such frustrations even as Bluente grows.

This conviction has also influenced how she thinks about the company’s pace of expansion — Tay is prepared to sacrifice velocity to preserve customer confidence.

“I’m willing to grow more deliberately if that is what it takes to keep our promises aligned with what the product can reliably deliver,” she says, before offering a rather sobering assessment of entrepreneurial priorities: “Trust is much harder to rebuild than a growth target is to revise.”

The advantage of asking

Trust, of course, is a particularly complicated currency for young entrepreneurs operating in environments where professional authority has traditionally accumulated alongside experience.

Tay recognises the difficulty immediately. “The hardest part is earning trust before you have a long track record to point to,” she concedes, particularly when customers handle documents whose contents carry commercial or legal consequences.

Yet she had encountered a version of this dilemma before founding Bluente. At Bain, she advised senior stakeholders with years of expertise in their fields. Rather than attempting to compensate for the age difference through exaggerated confidence, she developed an appreciation for preparation, judgement, and attentive listening.

Being young, she has discovered, also creates a peculiar opportunity. “It gives us the ‘excuse’ to ask for advice, ask for feedback,” she reflects, adding that such conversations can open doors to senior decision-makers willing to share their experience.

Still, curiosity alone cannot sustain credibility. As Tay understands it, the real advantage comes from demonstrating that the advice has been heard, considered, and translated into something useful.

The work nobody sees

Today, Tay’s longer-term ambition is to build what she describes as “the document infrastructure for the AI era”, allowing information to move between languages, enterprise systems, and AI applications without requiring people to reconstruct it repeatedly.

The proposition sounds expansive, although its success may ultimately depend on the smallest and least glamorous details. A correctly preserved table; a contract whose structure survives translation; a professional who can proceed directly to reviewing its contents rather than repairing the document first.

These are hardly the achievements that usually command attention in conversations about artificial intelligence. Yet they reveal why Tay has developed such a particular understanding of progress.

As she explains, the company chooses its technology according to what customers need, rather than what happens to be fashionable. For some problems, AI provides the answer. For others, traditional engineering offers greater certainty. And for decisions that require professional judgement, the responsibility must remain with people.

Perhaps the most persuasive evidence of Bluente’s success will eventually be how little attention its technology demands. A document arrives, the information remains intact, and somebody gets on with their work.

For a founder whose business began with the frustration of learning unfamiliar words, the ambition has grown considerably. She now wants to eliminate the work that language barriers leave behind.

And sometimes, achieving that requires knowing precisely when artificial intelligence should get out of the way.

Art direction: Fazlie Hashim
Photographer: Athirah Annissa
Stylist: Dolphin Yeo
Hair: Joanne Ong, using Kevin Murphy
Make-up: Keith Bryant Lee, using Clarins

For more stories on Strictly Zillennial, visit here.