Workera Adds AI Translation to Its Skills Assessments

A skills test given only in English doesn't just measure skills. It also measures English fluency. Workera, a Palo Alto, Calif.-based skills intelligence firm, thinks it's found a fix.

The company announced the launch of AI-First Translations today, which makes assessments, mentoring tools and admin features available in employees' own languages.

The Problem Hiding Inside the Test Score

In its announcement, Workera used research from Pearson to make its case: 92% of employees globally say English matters for getting ahead at work, yet only 7% of non-native speakers say they can actually communicate well on the job. That's a substantial gap.

With that in mind, the results of skills tests for non-native English speakers get muddy fast. Did an engineer in São Paulo fail because she didn't know the material, or because the questions were written in a language she wasn't fluent in? Right now, most companies can't tell the difference. That's the problem today's release aims to solve.

No Human Checks Workera's AI-First Translations

The translation happens automatically, meaning there's no human review. Workera treats the English version as the official one, labels translated content so users know it came from AI and lets employees flag mistakes so the system can improve. Workera says this approach lets companies roll out testing in new countries in weeks instead of months, because they don't have to wait on translators or for a review process.

"When teams across the globe are measured on real capability, your view of talent stops being a sample and becomes the whole picture," said Kian Katanforoosh, Workera's founder and CEO.

The feature is rolling out with the following languages, with more on the horizon, based on customer demand: English, Spanish (Spain), Portuguese (Brazil), Spanish (Mexico), French, Italian, German, Turkish, Korean, Simplified Chinese and Japanese. The announcement didn't include any accuracy data or explanations of the procedure for when an employee flags a bad translation — does a person look at it, or does the correction loop stay automated too? It does note that the translation layer "is built for rapid usability and comprehension rather than certified, word-perfect localization."

A Second Bet: Skip the Test Entirely

Translation solves one part of the "are we measuring the right thing" problem. Workera recently introduced Ambient in private preview, that approaches the problem from a different angle. The tool reads signals from the tools people work in — code they write, documents they produce, conversations they have — to build a picture of their skills without asking them to sit for an assessment.

The company's broader argument, laid out by Katanforoosh in an earlier Reworked interview, is that most companies don't know what their employees can do. Job titles and resumes go stale fast, and self-reported skills are unreliable. Workera has also promoted an idea it calls "learning agility," or how fast someone can turn new knowledge into results, which it argues matters more for career growth than tenure or credentials.

Have a tip to share with our editorial team? Drop us a line: