On a Telegram message, fetches and embeds news articles, then answers with an OpenAI agent over vector search and replies in Telegram.
When files are added to Google Drive, indexes them into Milvus with Cohere embeddings so an OpenAI chat agent answers questions.
On a WhatsApp message, an OpenAI agent answers using an in-memory vector store with conversation memory and replies via WhatsApp.
On a chat message or Google Drive upload, embeds files into Supabase and answers questions with an OpenAI RAG agent.