Project story
NewsGenie addresses LLM knowledge cutoff limitations through graph-based multi-tool routing. The system analyzes user intent to either answer from training data, query categorized news headlines via NewsAPI, or perform live web research via Tavily with structured error handling.
Business outcome
Demonstrated production agentic patterns (cyclical tool routing, state management, fallback recovery) and delivered an interactive Streamlit UI for live queries.
Technology stack
- LangGraph
- LangChain
- Google Gemini
- NewsAPI
- Tavily Search
- Streamlit
- Azure App Service