Artificial intelligence may eventually become the perfect universal translator, but even though machine translation may be lightning fast, it still falls behind the accuracy of human translators.

Portuguese startup Unbabel believes its “AI-powered, human-refined” translations combine the best of both worlds. Today it announced it has raised $23 million in Series B funding to develop its machine translation platform further.

The funding round was led by Scale Venture Partners, and it included participation from Microsoft Ventures, Salesforce Ventures, Samsung Next, Notion Capital, Caixa Capital and Funders Club. Unbabel has raised more than $29.5 million to date.

Founded in 2013 in Lisbon, Portugal, Unbabel uses AI to do the heavy lifting for its translations, but then it also has a network of more than 50,000 human translators who check over the AI’s work to make sure that it sounds natural and not robotic. Unbabel’s Translation API allows businesses to integrate its service with their own platforms, and the startup also offers direct support for Salesforce, Zendesk, Freshdesk and others.

Unbabel translates text from dozens of languages, including French, Japanese, Korean, Russian, both to and from English. It also offers translations between a few non-English languages, such as translating Italian to German or French to Spanish.

“We started this business with the ambition to reach the 70 percent of internet users who don’t speak English, even as a second language,” Unbabel co-founder and Chief Executive Vasco Pedro (pictured, far left) said in a statement. “But the demand for global brands to serve customers in their native language has accelerated so fast that Unbabel must now take the next step of scale to meet it. In an Unbabel-powered internet, every business will be able to go global from day one.”

Andy Vitus, partner at Scale Venture Partners, added that Unbabel is eliminating the language barrier for large enterprises to serve customers worldwide. “Never before have companies been able to easily integrate translation into daily workflow in a cost effective way,” he said. “This integration is setting up Unbabel to be the central hub for translation in the enterprise.”

Besides improving its machine learning systems, Unbabel said it will use the new funding to hire more people in sales and marketing.

Photo: Unbabel