Generative AI startups like Eleven Labs have garnered attention for their voice translation capabilities. However, the demand for content translation remains robust, especially for companies with global operations. This market has seen significant investment, with companies like Unbabel in Portugal raising $106 million. Since 2010, EasyTranslate has carved a niche in content translation, leveraging machine learning to match content with suitable freelance translators. Now, EasyTranslate is taking a bold step with its new AI-driven platform, “HumanAI,” which promises to revolutionize the translation process.
EasyTranslate has shifted its business model from traditional human-based services to becoming an AI technology provider. According to founder Frederik R. Pedersen, this transition aims to reduce costs and accelerate the translation process. Unlike typical translation services that rely on human editing of machine-translated content, HumanAI integrates large language models (LLMs) and utilizes short-term memory in the LLMs for more accurate translations. Humans are only involved when necessary, significantly cutting down translation times and costs.
How HumanAI Works
HumanAI combines LLMs, including OpenAI’s models, with EasyTranslate’s proprietary recommendation systems. By incorporating its algorithms and customer data, the platform offers tailored content translations. The key innovation lies in using LLMs to generate short-term memory, enabling the platform to translate content with context-specific accuracy. This method, known as retrieval augmented generation, vectors content into a database, performs semantic searches, and uses the findings to create a short-term memory for the LLM.
This approach allows HumanAI to handle different types of English, such as marketing copy or financial reports, while preserving the intended meaning. Pedersen explains that traditional neural machine translation engines are combined with customer-specific data to create a foundation for accurate localization and translation.
HumanAI significantly reduces translation costs, claiming a 90% reduction, and prices its services at €0.01 per translated word. Its clientele includes global companies like Wix and Monday.com. The platform’s ability to provide contextually accurate translations is crucial for businesses with extensive content needing translation. For instance, companies like Adobe invest heavily in ensuring their terminologies align across markets, highlighting the importance of precise and localized translations.
Pedersen emphasizes that EasyTranslate’s goal is not to become a pure AI service but to combine AI with human expertise to add value for customers. While AI can handle a large volume of content, human intervention is essential for fine-tuning and ensuring the translations sound right. By identifying parts of the content with low confidence scores and using humans to correct them, HumanAI enhances productivity.
The Future of Translation Services
EasyTranslate’s approach of blending AI with human feedback sets it apart from purely AI-based solutions. As language evolves, human oversight remains vital to control and improve AI models. EasyTranslate aims to maintain this balance to offer superior translation services.
EasyTranslate has raised €3 million, backed by private equity, debt financing, angel investors in Copenhagen, and the Danish Innovation Fund. This financial backing supports its mission to innovate and lead in the translation service industry.
In summary, EasyTranslate’s HumanAI platform represents a significant advancement in the translation industry by combining AI’s efficiency with human expertise to deliver accurate, context-specific translations. This hybrid approach positions EasyTranslate to compete effectively in a market increasingly dominated by AI-based solutions.
See also: CData Secures $350M To Enhance Data Integration For AI Models
