The most important part of building an app with AI happens before you ask AI to build it. And most people skip it. When I'm working in tools like Lovable, Base44 or other AI builders, I don't want to jump straight into Build mode. I want to spend time in Plan mode first. Before building anything, I work through: → What problem are we solving? → Who are the users? → What are they trying to accomplish? → What are the requirements? → What entities and data do we need? → How do they relate to each other? → What are the core workflows? → What states can things be in? → What happens when something goes wrong? → What actually belongs in the MVP? Then I ask AI to challenge the plan. What's missing? What doesn't make sense? What edge cases have we overlooked? What assumptions are we making? Only after we've worked through those questions do I want to build. Because AI is extremely good at doing something you didn't actually ask for. It can build the wrong product incredibly quickly. The better the product structure you give it, the more useful AI becomes. That's why I think the future of AI-assisted product building isn't just about getting better at prompting AI to build. It's about getting better at prompting AI to think with you before you build. My workflow is increasingly: Plan → Challenge → Refine → Build → Test → Iterate That's the approach I'm using while building products like Anywhere Pilates Studio Manager. And it's changed the way I think about AI-assisted product development. Don't just tell AI what to build. Figure out what should be built first. Building with Base44