We've just seen a massive leap forward with a major AI translation tool called Lara Translate, developed by the folks at Translated. They've just expanded support to cover the 32 languages spoken by the most people on the planet, bringing the total reach to a staggering 2.8 billion native speakers. This isn't just adding a few more languages; it's a strategic move that includes crucial ones like Arabic, Hebrew, Japanese, Korean, Thai, Malay, and Indonesian, really opening up huge markets. But the big story isn't just the number of languages; it's how this changes the game for translation teams. We're talking about a fundamental shift in how organisations handle multilingual content, making it possible to translate, share, and scale like never before, all while keeping that essential human touch.
One of the key innovations here is the new Team plan. This is brilliant because it lets companies build a shared AI model that's specifically trained on their own translation memories and language assets. Think about that – you can fine-tune the AI to perfectly capture your company's unique tone, terminology, and overall intent. This means whether you're localising advertising copy, drafting internal documents, or rolling out global marketing campaigns, every piece of content speaks with the same consistent voice across all departments. And getting started sounds pretty straightforward too. You don't even need to be an existing customer; you can just upload your existing language data or get help generating it from previously approved materials. Crucially, your data stays private, exclusively used for training your model and not feeding into anyone else's system.
Hand-in-hand with this expansion is the launch of the Lara Agent. This is essentially a generative AI assistant specifically built to automate localisation workflows. Imagine being able to simply ask the agent to "Translate this file into Tier 1 languages," or "Adapt the model using this new dataset," or even "Create a JSON file for development." It’s designed to handle these common, often time-consuming tasks automatically. The agent is already integrated with popular platforms like Claude Desktop and Cursor, with more integrations planned, and it's part of this new Model Context Protocol – a shared standard allowing these AI agents to work together across different environments. The initial response has been incredible; the Lara Agent saw more downloads via the MCP in just 24 hours than Translated's API and SDK combined did in two months! That tells you there's a real appetite for this kind of workflow automation.
For larger organisations dealing with high volumes of content, there's a deep integration with TranslationOS, which is Translated’s AI-first localisation platform. This, combined with a new connector specifically for Contentful, allows users to push content straight from their content management system for immediate translation and then bring it back without any manual exporting or frustrating formatting issues. Global players like Skyscanner and Asana are already using this setup, significantly streamlining their multilingual content pipelines and ensuring updates are faster and much more consistent across all supported languages.
And it's important to remember that this isn't just a new product launch; it's built on a solid foundation of ongoing research. The team behind this is actively pushing the boundaries of what language AI can do, presenting their latest work on deep learning and scalable reasoning at major conferences. They were at ICLR 2025 recently and are joining the NLP community at NAACL 2025 this month, continuing to contribute to and shape how AI understands and communicates across languages. It really feels like the industry is moving rapidly towards more integrated, AI-powered workflows, and this announcement is a significant step in that direction.