Jason Liang

One of the biggest barriers enterprises face in scaling generative and agentic AI is ensuring accuracy, reliability, and trustworthiness at every stage of the lifecycle. At the root of this is often a lack of understanding and oversight of the data used.

This session explores how human-in-the-loop evaluation bridges the gap between a proof of concept and a production-grade AI system. We’ll explore how leading companies design evaluation programs that combine a deep understanding of data, human expertise, and automation to improve performance and maintain trust.

You’ll learn:

• Why most evaluations fail and where the biggest opportunities for improvement lie.
• How human-in-the-loop evaluation drives measurable gains in accuracy and reliability.
• How to optimize your LLM judge to work reliably and at scale.
• How to structure a data-driven evaluation program that scales with your AI initiatives.
• How leading companies bring AI from experimentation to production with continuous evaluation.