Build an AI Coding Agent That Works While You Sleep (Live Demo)

Watch a live build of a self-correcting AI coding agent using Ralph Loops — an agentic workflow pattern that plans, executes, tests, and fixes its own code until the job is done. In this session, Udacity's AI Curriculum Lead Val Scarlata builds a real executive assistant app that connects to Gmail and Google Calendar. She walks through how to write a product spec (PRD) with user stories and acceptance criteria that an AI agent can actually act on, then runs the Ralph Loop — a bash script that kicks off Claude Code in an autonomous loop — live on camera. You'll learn: -How to break a project into agent-ready user stories -How Ralph Loops handle long-running tasks and context management -When this pattern makes sense (and when it doesn't) -What it actually looks like when AI writes, tests, and commits its own code 🚀 Take Your AI Skills to the Next Level Check our our GitHub repo: https://github.com/udacity/agentic-loop-webinar Download our Field Guide to Ralph Loops: https://www.udacity.com/resources/ralph-loops-field-guide Go beyond single chatbots to engineer sophisticated, coordinated teams of AI agents in our Agentic AI Nanodegree Program ➡️ https://www.udacity.com/course/agentic-ai--nd900 #agenticai #artificialintelligence #machinelearning #llm #aiagents #productionai #techwebinar #udacity