Users expect answers in their language, even when your content was never written for them. In this session, we’ll explore two common patterns for multilingual AI systems on Azure: one where documents are available in each language and one where the corpus exists in only a single language. Through live demos, we’ll compare locale-aware retrieval, multilingual search, document translation, and translation after retrieval, and show when an LLM responding in the user’s language is enough—and when the retrieval architecture has to do more of the work.

You will learn:

• About designing AI systems for a global audience
• About embedding and retrieval strategies for multi-lingual retrieval
• About translation services