MotherDuck has introduced the Answering Machine, a tool that allows users to query their data using natural language through platforms like Claude, Gemini, or ChatGPT. This setup requires no SQL knowledge; users simply add a connector endpoint to their LLM client and can start asking questions. The author initially approached the tool skeptically but was impressed when it returned accurate insights about business tier users and revenue just minutes after setup, demonstrating its applicability to messy real-world data. During a test, the author queried how customer usage varies by day of the week and the impact of US holidays. The tool navigated through the database, analyzed patterns, and provided detailed insights, including a 20% drop in usage on weekends and a surprising increase in activity during Thanksgiving week. The analysis revealed that while overall holiday usage dropped by 8.6%, paying users showed virtually zero impact due to automated data processes. Claude not only reported the findings but also speculated on the reasons behind them, such as the dominance of automated workloads among paying customers. The piece reflects on the evolution of natural language queries in data analytics. Previous attempts, like those from ThoughtSpot, faced challenges making natural language processing work effectively in large and complex datasets. With the advent of large language models (LLMs) like ChatGPT, the ability to convert natural language into SQL queries has improved significantly. However, the author notes that while LLMs can generate decent results, they still struggle with accuracy when it comes to critical business decisions, emphasizing the need for a robust solution in self-service analytics.