Understanding and testing LLM output is becoming an art-form. Current methods to improve LLM prompt outputs has suffered from the inability to systematically test outputs. To get towards a data driven prompt testing methodology, a framework is required. Since prompts are the foundation of AI agents, prompt testing methods are essential.
This presentation will focus on testing and test driven prompt engineering. We will discuss the state of prompt engineering, current AI prompt testing methods, available testing frameworks, and a deep dive into promptfoo.