The dishwasher runs a full cycle every night. I have no idea what’s happening inside it. I don’t know the exact temperature of the water at any point, I don’t know how it sequences the spray arms, and I have never once thought about any of this before putting a mug on the top rack and pressing start. The dishes come out clean. That’s the whole deal.
AI works the same way. And if you’ve been holding off because you think you need to understand it first, that’s the belief this post is specifically here to dislodge.
The Understanding Trap
There’s a version of “getting into AI” that looks like this: watching explainer videos, reading about neural networks, trying to understand how large language models actually process text. That path exists, it’s genuinely interesting, and it is completely unnecessary if what you want is to build things with AI.
I’ve never written production code in any job I’ve held. Not once. After 25 years in sales and training, I know how to describe a problem clearly, how to spot when a solution doesn’t actually solve it, and how to iterate until it does. That’s what I bring to building. Claude Code handles the rest. The apps are real, the database is real, the checkout page is real. None of it required me to understand what is happening at the code level any more than I need to understand the spray arm sequence.
The prerequisite for using AI isn’t comprehension. It’s clarity about what you want.
What You Actually Need Before Anything Else
This is the foundational piece that I think most people skip, and it’s why they stall. They sit down with a blank AI chat window and type something vague, get something vague back, and conclude that AI isn’t as useful as advertised.
The skill that makes everything else work is the ability to describe your problem in plain, specific language. Not technical language. Plain language. “I want a dashboard that shows me which of my content ideas still don’t have a hook written for them” is a real instruction I’ve given. It’s a sentence anyone could write. What came back was a working tool.
The same logic applies to the Voice to Blog Post workflow I packaged up as a $7 product on Pinball. The workflow exists because I could describe exactly what I wanted: I talk, the transcript gets cleaned up, it gets shaped into a blog post in my voice, and I’m done in five minutes. I didn’t build that by understanding AI. I built it by knowing what result I needed and being specific enough that Claude could execute.
The Stack Doesn’t Need to Be Complicated
One thing that trips people up is thinking they need to make a lot of decisions about tools before they can start. You don’t. My whole build stack is about five things: Claude Code as the agent that writes my code, Convex as the database, the Anthropic API or Gemini API depending on what the app needs, and Netlify to host whatever gets built. That’s it.
I didn’t pick those tools because I evaluated seventeen alternatives. I picked them because I started somewhere, something worked, and I kept using it. The Recipe App I built runs on that exact stack. It reads a photo of a handwritten recipe card, imports from URLs, takes voice memos, and lets you talk to an AI about what to make for dinner. I could not build that if I had to write the code myself. But I could describe it. And I can describe exactly which part isn’t working when something goes wrong, which is how iteration actually happens.
You own the workflow when you build it this way. No monthly subscription to a SaaS company that decides next quarter to restructure their pricing. No waiting for a developer to have availability. You describe, it builds, you adjust.
The Only Prerequisite That Actually Matters
Before any tutorial, any walkthrough, any specific tool recommendation I’ll point you toward, this is the thing that has to land first: you don’t need to understand the technology. You need to understand your problem.
That’s it. If you can describe what’s currently taking you too long, costing you money, or falling through the cracks because there’s no system for it, you have everything you need to start. The AI fills in everything else. You stay in the seat of someone who knows what they want and can tell when they’re not getting it.
I’m going to walk you through setting up exactly the kind of build-and-iterate process I use. But none of that’s useful until this one thing clicks: you’ve been the right person to do this all along. You just didn’t know the dishwasher didn’t need you to understand it.
Pick one problem you have right now, something that happens repeatedly and costs you time. Write it out in two or three sentences, as if you were explaining it to someone who just asked you what’s frustrating about your week. That description is your first prompt. You’re already further along than you think.