Eval Driven Development: Why you need it to make your Agentic Application survive Production

Agentic systems unlock capabilities that traditional “deterministic” applications simply can’t deliver. But there’s a catch: their probabilistic nature introduces real and often unexpected risks—hallucinations, context drift, prompt degradation, and multi-step workflows that evolve in surprising ways. And you definitely don’t want to discover those in production.

The good news: there is a remedy. We can borrow the most reliable practice from deterministic software—test-driven development—and adapt it to the agentic world. The result is Eval-Driven Development (EDD): a systematic, engineering-first approach to bringing reliability into inherently probabilistic systems.

In this talk, we explore Eval-Driven Agentic Development and how it can transform your agents from unpredictable to reliable - using the (only?) full-fledged Java/Kotlin Eval framework: Dokimos. We’ll dive into techniques, tools, and patterns needed to make evals a first-class citizen of your development process—demonstrating all of these in a real-world application built with AI frameworks like Spring-AI, LanchChain4J, and Koog.

Along the way, you will learn how to:

• Design layered eval suites
• Create golden datasets
• Define rubrics and scoring criteria
• Extract eval signals from agent traces
• Use regression gates to detect degraded behavior after changes
• Build a continuous eval loop

Evaluation-Driven Development is the only path forward for reliable, smart next-generation applications. Let's learn how to walk it - together!