Evaluating the agentic platform: Building evaluation infrastructure for agents across the entire SDLC

Evaluating a single AI agent is a solved problem. Evaluating agents across every stage of your SDLC, where outputs feed inputs and failures compound, is a platform engineering challenge. This workshop covers how to build evaluation infrastructure that works end-to-end: what to measure at each stage, how to connect signals across the pipeline, and how to make eval a first-class part of your agentic platform rather than an afterthought end-to-end evaluation infrastructure.