ساخت ربات پشتیبانی مشتری تلگرام با هوش مصنوعی RAG و ارجاع به اپراتور (Human Escalation)

The video walks through building a Telegram customer support bot that uses Retrieval-Augmented Generation (RAG) and can hand off conversations to a human operator. It covers the architecture, setting up Telegram triggers, connecting to a Supabase vector store, using OpenAI embeddings, implementing escalation logic, storing conversations in a window buffer memory, and updating reports in a sheet or database. Key timestamps: 0:00 – introduction, 0:54 – demo, 2:11 – escalation test, 4:08 – start of training, 5:03 – prompt settings, 5:48 – memory and Google Sheet, 6:16 – escalation implementation, 7:01 – RAG system, 7:23 – Supabase configuration, 8:08 – knowledge base prep, 9:19 – Supabase DB setup, 10:14 – data ingestion, 11:21 – final RAG connection, 12:47 – review of logged data, 13:36 – summary.