I built a simple RAG-style chatbot with ASP.NET Core MVC and the Gemini API that lets a user upload a PDF and ask questions about it. The app extracts the PDF's text, sends it with the question to Gemini using a prompt that restricts answers to the PDF's content only, and displays the result. I used a PDF I generated on Backend Development with .NET and tested it with 10 prepared questions plus one unrelated one to confirm it doesn't invent answers.