Crowdin

A few months ago, I went further into implementing 𝗻𝗲𝘄 𝗮𝗻𝗱 𝗺𝗼𝗿𝗲 𝗲𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝘁 𝘄𝗮𝘆𝘀 𝘁𝗼 𝘁𝗿𝗮𝗻𝘀𝗹𝗮𝘁𝗲 𝗮𝘀𝘀𝗲𝘁𝘀 𝗮𝗻𝗱 𝘀𝘁𝗿𝗲𝗮𝗺𝗹𝗶𝗻𝗲 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 for Sinch, where I work as #LocalizationManager. We already rely on 𝗲𝘅𝗰𝗲𝗹𝗹𝗲𝗻𝘁 𝘁𝗼𝗼𝗹𝘀 like DeepL, Blackbird.io, and the solutions provided by Smartling, but we wanted to go beyond that. 💡 Together with the amazing Nicola Calabrese from Undertow, we designed the 𝗻𝗲𝘅𝘁 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝗼𝗻 𝗼𝗳 𝘁𝗵𝗲 𝗦𝗶𝗻𝗰𝗵 𝗹𝗼𝗰𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗽𝗿𝗼𝗴𝗿𝗮𝗺, one that combines Large Language Models (#LLMs) with proven localization technologies. 💻 Our goal: to create a scalable, intelligent system that adapts to content type, language, and business impact. To make that happen, we have: ✅ 𝗗𝗲𝘀𝗶𝗴𝗻𝗲𝗱 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗲𝗱 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 (that for example include term extraction and glossary creation with context and equivalent term proposals in our target languages) ✅ 𝗧𝗮𝗶𝗹𝗼𝗿𝗲𝗱 𝗽𝗿𝗼𝗰𝗲𝘀𝘀𝗲𝘀 based on 𝗰𝗼𝗻𝘁𝗲𝗻𝘁 𝘁𝘆𝗽𝗲, 𝘁𝗮𝗿𝗴𝗲𝘁 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲, 𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲 𝗺𝗲𝘁𝗿𝗶𝗰𝘀 (traffic, asset life, etc.)... ✅ 𝗦𝘂𝗰𝗰𝗲𝘀𝘀𝗳𝘂𝗹𝗹𝘆 𝘁𝗲𝘀𝘁𝗲𝗱 𝘁𝗵𝗲 𝘂𝘀𝗲 𝗼𝗳 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝗟𝗟𝗠𝘀 𝗮𝘁 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝘀𝘁𝗲𝗽𝘀 ( translator, reviewer, QA assistant, and more) all within the Crowdin AI environment, achieving excellent results (suprisingly for me, the best being using Gemini as both #translator and #reviewer) 💪 The outcome? A smarter, faster, and more cost-effective localization pipeline that brings together the best of AI innovation and traditional localization expertise. ⚙️✨ If you’re curious about how we’re reshaping localization to help companies scale intelligently and sustainably, I’d love to connect, or you can schedule a chat here 👉 https://lnkd.in/dKBhrN34 #Localization #AI #Translation #Innovation #LLMs #Automation #CrowdinAI #SmartGrowth #Sinch

AI as MT works, but only when you provide context. That's why I tested Crowdin's Vector Cloud. It lets AI learn from your own translated content, documents, and style guides. Many thanks to Crowdin for sponsoring this video. 👇 Links to the Nimdzi Insights report I mentioned and Vector Cloud are in the comments. You can also watch it on YouTube or Substack. #localization #tms #vector-database #ai #translation #crowdin #openai #ad

Now There Is A Smarter Way Than TM To Leverage Your Past Translation Work 🔥🔥🔥⬇️ Could it be that, finally, we are no longer bound by the rigid structure of bilingual source-target string pairs? In this clip from Episode 4 of the Crowdin Power Up Series, Dorota Pawlak and I discuss the platform's new Vector Cloud functionality, which lets you teach the AI your style, tone, and terminology simply by uploading your past work, monolingual or bilingual. Style guides, PDFs, manuals, even old projects become rich reference material. The result? Translations that actually sound like you. We believe this could well open the door to a new kind of value delivery, one where creativity and strategic differentiation are built into the process. It will certainly make it much easier to produce different versions of your translated copy. Could this be the future of multilingual copywriting? Do you think it could be valuable for your clients to have a choice between different styles of translation? With Crowdin’s new Vector Cloud functionality, that kind of creative flexibility might just have leveled up. Instead of delivering a single version and hoping it sticks, translators can now use their style guides, brand tone samples, and prior projects to generate multiple options, each aligned with client-specific context. Marketing teams are already used to A/B testing copy. Why not give them the same flexibility in translated content? Imagine offering three subtly different translations for a project, each tailored to a specific voice, playful, authoritative, or aspirational, and letting the client choose what best fits their campaign. This could be about creativity as much as about strategy. The AI isn’t guessing anymore, it’s reading your reference files, internalizing brand tone, and translating with nuance. What would it mean for your business if your translations came with options your client didn’t even know they needed? Would your clients appreciate the choice, you think? Check out the full Episode 4 of the Crowdin Power Up Series to find out more about how this works in practice. Let us know whether you would use this in your client work.