We created an AI-powered conversational bot designed to efficiently gather and present current news on finance, economy, and politics from YouTube news channels. Users can inquire about the latest economic news and receive comprehensive answers through a Retrieval-Augmented Generation (RAG) system. This system not only provides relevant information but also analyzes and displays the sentiment associated with each context of the query. Additionally, it asses public sentiments by analyzing YouTube comments related to the news topics.
We downloaded 221 videos as audio files spread over the course of 3 days from 10 YouTube channels including but not limited to Yahoo finance, Bloomberg television, The Economist, Financial Times, Reuters, The Wall Street Journal, Washington Post.
Core technologies used: YouTube Data API v3, Whisper OpenAi, Pinecone vector database, LLMs, RAG framework and Langchain