Course notes from 'Building Agentic AI Applications with a Problem-First Approach' - moving beyond clever demos toward measurable, reliable agent systems with practical playbooks for problem-first framing, early evaluations, guardrails as product requirements, and iteration with data.