This week's newsletter reveals why AI sometimes makes things up and how to get dramatically better, more reliable answers from any AI assistant. Learn the simple conversation techniques that reduce hallucinations, like asking for web searches, providing specific context, and letting AI ask clarifying questions. Plus, discover the latest open-source AI releases, including DeepSeek V3.1's hybrid thinking model, specialized tools for document conversion and image editing, and why smaller, focused models might be smarter than massive do-everything alternatives.