The author has been extensively using Claude Code for work, developing significant parts of LernerPython.com and integrating an AI tutor into courses and newsletters. This has led to writing less code manually while producing more and enjoying the process. A key takeaway is that AI agents like Claude Code generally do what you ask, but developers must be vigilant about potential gaps between intent, specification, and implementation. To effectively manage AI-generated code, one must implement rigorous validation, checklists, and follow-up questions within a controlled development environment. Four practices have significantly changed the author's workflow: a CLAUDE.md file acting as a contract for AI instructions, prioritizing tests over simple code differences, implementing detailed telemetry and logging, and relying on small commits as a rollback strategy. The author is also offering three workshops on using Claude Code with Python and Pandas, covering introductory and advanced topics. These workshops are hands-on and aim to teach practical application of AI in coding. An informational session is available for those considering the workshops.