Watch the event recording below

AI can write code. But it still can't understand people, and that's the part of software engineering that actually matters.

In this AI Exchange talk, Elena Vilimaite from WeBuild-AI shares what happened when she pointed an agentic coding pipeline at her own codebase and stepped away from the keyboard. The Ralph Wiggum loop is deceptively simple: give an agent a plan, then let it cycle, implement, review, fix, on repeat.

Elena walks through three real runs on her own code: a 24-minute one-shot that shipped subtle bugs, and a two-and-a-half-hour run where the agents caught and fixed their own mistakes. You'll see what made the difference (spoiler: it wasn't the loop), and why the human still isn't going anywhere just yet.

The true job of a software engineer, Elena argues, is turning human pain points into requirements, and someone still needs to review the massive pull requests agents are producing now.

Whether you're a developer exploring agentic workflows or a leader thinking about what AI means for your engineering team, this session offers an honest, practical look at cutting code AFK without losing the human in the loop.

About the speaker:

Elena is a Lead Software Engineer at WeBuild-AI.

Elena builds production AI systems and agentic development workflows for enterprise clients.

Her work focuses on the gap between AI-generated code and real software engineering, turning human pain points into requirements, and knowing when to step in and review what the agents produce.