The Evolution of Prompting: From Artisan Craft to Algorithmic Optimization

LLMs unlocked advanced capabilities for users of all technical abilities precisely because of their natural language interface. Building enterprise solutions involved prompt engineering - a process of trial and error. This talk explores how automated prompt optimization represents a paradigm shift in how we build high quality, maintainable enterprise solutions with LLMs. We'll examine the core limitations of hand-crafted prompting, then dive into how prompt optimizers solve these problems. The session will cover the current state of the art in prompt optimization, including the latest algorithms such as GEPA. We'll demonstrate measurable performance gains that prompt optimizers achieve over manual engineering - often discovering counterintuitive prompts that humans would never write but dramatically outperform our best handcrafted attempts. You'll leave understanding how prompt optimizers work and why they are essential for building high quality, maintainable GenAI solutions.