One habit has saved me countless hours when building AI systems. I don't start with code. I start with a blank page. Before I think about Claude Code, APIs, automations, or AI agents, I ask myself a few simple questions: - What kicks off this process? - What information is needed at each step? - Where does the work slow down? - Which decisions are repetitive, and which actually require human judgment? Only after those questions are answered do I start building. I've found that many automation projects become complicated because the workflow was never fully understood in the first place. The AI isn't confused. The process is. Once the workflow is clear, choosing the technology becomes much easier. Sometimes the solution is an AI agent. Sometimes it's a simple API integration. Sometimes it's just removing an unnecessary step that no one questioned before. The more systems I build, the more I realize this: Good software doesn't create good processes. Good processes make good software possible.