Building a Multilingual Multi-Agent Chat Application Using LangGraph — Part I

Building a Multilingual Agentic Chat Application Introduction: This project aims to address language barriers in the workplace by developing a multilingual chat application with an AI assistant named Aya. High-Level Framework: The application uses LangGraph to manage complex message flows, including AI assistant interactions. User Agents: Each user has an agent responsible for translating messages into their preferred language. Aya Agents: Various agents handle specific tasks within Aya, including question answering, documentation retrieval, summarization, and translation. AyaSupervisor Agent: Coordinates Aya agents and ensures only final messages reach users. AyaSummarizer Agent: Determines the number of messages to summarize and identifies action items. AyaTranslator Agent: Translates incoming messages to English for internal Aya processing. Prompt Design: Custom prompts guide the LLM to generate consistent responses in specific formats. Deployment: The application is deployed using FastAPI and a web UI for user interaction.