Kenny Warner on How AI Models Really Work, Why Clean Data Wins, and the Future of Restaurant AI

The conversation centers on the integration of data science and artificial intelligence within the culinary industry, specifically how advanced systems transform raw inputs like recipes into actionable business intelligence. Kenny Warner, head of data science and AI at Meez, explains the mechanics of Large Language Models (LLMs), detailing how embedding, attention mechanisms, and fine-tuning allow models to process specific, proprietary data sets accurately. The discussion highlights the shift from traditional data storage to dynamic transformation, where systems automatically calculate costs, yields, and margins. Furthermore, the participants examine the practical challenges of data infrastructure in restaurants, the utility of Model Context Protocol (MCP) for API-driven data retrieval, and the necessity of maintaining clean, validated data to avoid hallucinations in AI-driven decision-making. The dialogue concludes with a segment on trivia, underscoring the intersection of technical expertise and lighthearted engagement.