video1.0<iframe src="https://www.loom.com/embed/e439e796d61546a680d9a1a0962460ab" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/sessions/thumbnails/e439e796d61546a680d9a1a0962460ab-94ee6d7e9f280b3e.gif328.933In my final Spring semester CodePath project, I upgraded my initial rule based pet care scheduler into a full applied AI system. The workflow uses RAG to pull pet care rules and an agent workflow to analyze, plan, call the API, evaluate safety guardrails, and then route for human review to accept or revise. I ran a test evaluator, and the results show passed or not passed checks for violations like how many rooms. In the demo, Bella and her schedule used 19 relevant safety rules, Bulldog used 16, and all three tests were passed. No action was explicitly requested from viewers.