You may think that new generative AI startups like Eleven laboratories are the hottest market for translation services. But language translation was long preceded by another market that startups started focusing on some time ago: content translation. Any company with an international presence needs to have its content translated worldwide, so this remains a big market. This has been demonstrated by the $106 million raised so far by companies like Unbabel in Portugal (which last 60 million US dollars raised).

EasyTranslatewhich specializes in content translation, has been around since 2010. It uses machine learning models to figure out which freelance translators are best suited to translate certain types of content. But now the company is taking a new direction with a new, generative AI-driven platform called “HumanAI.”

“We changed the entire business model from a human services-based business model to an AI technology provider to reduce costs and speed up the process,” company founder Frederik R. Pedersen told TechCrunch.

Most translation services offer machine-translated content, a small portion of which is edited by humans. However, translators often have to review the entire machine-generated translation to understand the context and make sense of the content. EasyTranslate’s HumanAI platform turns this on its head: it takes content, combines it with large language models (LLMs), and leverages the short-term memory in the LLM to translate content more accurately. What’s more, humans are only involved where necessary, reducing translation time and costs.

To do this, HumanAI uses a mix of LLMs, including the one offered by OpenAI, as well as its own recommendation systems. The platform uses its own algorithms and customer data to provide tailored content translations.

The secret of the pivot, says Pedersen, lies in using LLMs to create short-term memory, so that the platform can read a translation in general English and convert it into specific English. It “vectorizes” content into a database, which allows it to perform a semantic search and find similarities between content, which are then used to create short-term memory with an LLM (this is also known as Retrieval extended generation).

This means that the platform can use any number of LLMs to translate, for example, between English in marketing texts and English in financial reports, while always preserving the meaning of the text.

“We can combine the more traditional neural machine translation engines with customer-specific data to create a foundation for the localization and translation process. For example, we can move from generic language to customer-specific language,” he said.

Why is this important? Pedersen explains: “You can get a grammatically correct machine translation, but it still doesn’t sound right. So we identify which part of the content has a low confidence score and then have humans correct it. This combination increases our productivity enormously.”

Pederson claimed that HumanAI can reduce translation costs by 90% and calculates that its services cost €0.01 per word translated. Its clients include global companies such as Wix and Monday.com.

And pricing is a particularly difficult puzzle in this area because companies have a large amount of content that needs to be translated.

“At Adobe, there’s a whole team dedicated to how terminologies align across markets. And when we look at global brands, there’s a lot of effort put into making sure they’re perceived correctly locally,” Pedersen said.

The question, however, is how can EasyTranslate compete against pure AI-based solutions, which are likely to get better over time?

“Our goal is not to become a pure AI [service]”I think our goal is to create the added value that comes with combining humans and AI and offer that service to customers. AI still needs human feedback to improve,” he said.

“It’s one thing to say you want to do all the content creation and all the translations yourself, but it’s another to make sure you can actually control the model. You need to have some people controlling the models because people are not machines and language is constantly changing.”

EasyTranslate has raised a total of €3 million so far and is supported by private equity, debt, some angel investors in Copenhagen and the Danish Innovation Fund.