"In the world, there are more than 7000 languages, but only about 200 of them have online translation tools, leaving hundreds of millions of people worldwide without the ability to communicate easily." Hello everybody, Previously I already posted on PH about Gaia, a no-code platform to train and deploy neural machine translation models, and while it indeed lowered the barrier for non-technical people to build on it, the issue of data remained with tens of thousands of parallel sentences needed to build a translation model between new language pairs. Well, now Gaia fixes that also. I am happy to introduce "Translation agents", LLMs combined with retrieval mechanisms that show superior performance, even when compared to NLLB-200 from Meta (Check the technical details here: https://gaia-ml.com/documentation). You can start building translation agents for low-resource languages with as few as 5000 parallel sentences. What is more exciting, personally, is that students, researchers, and college professors in Peru have already build translation agents for 7 languages in Peru that previosuly had no translation technology, check them here: https://gaia-ml.com/models