Your App's Brain

In the Quickstart you set up your builder, Claude Code, the contractor who does the construction. Now meet the second AI, the one that confuses almost everyone: your app’s brain.

Here’s what the brain actually is. When a real user opens your finished trip planner and types “five days in Tokyo,” something has to invent that itinerary on the spot. It doesn’t exist anywhere yet. There’s no list of pre-written Tokyo trips waiting in a drawer. Some intelligence has to read the request and write back a real, specific, day-by-day plan. That intelligence is an AI service your app calls over the internet, and it’s the brain inside your product. Your app sends it the question, the brain sends back the answer, your app shows it to the user.

Now here’s the part nobody tells you up front, and it’s the whole reason this is its own lesson: getting a brain is a product decision, not a setup chore. Your builder, you absolutely need, it’s how you make anything at all. But the brain? That depends entirely on what you’re building. For some apps the brain is the entire point. For others it’s optional, or something you add much later, or something you never need. So before you sign up for anything or spend a cent, let’s figure out which kind of app you have.

This lesson does two things. First, it helps you decide whether your app needs a brain at all. Second, if it does, it gets you a key so your app can talk to one. By the end, you’ll know exactly what you’re paying for, and you’ll have what you need and nothing more.

Try this. Describe your app in one sentence, the way you’d say it to a friend. Then look at the verbs.

If your sentence has words like generates, writes, suggests, creates, answers, summarizes, or plans in it, your app needs a brain. That’s the part doing the thinking, and the thinking has to come from somewhere. Our trip planner’s sentence is “you type a destination and AI plans your trip day by day.” The planning is the brain. Take the brain away and you’re left with an empty page and a button that does nothing.

But not every app is like that. Plenty of real, useful apps don’t generate anything, they store and organize what people already have. A shared shopping list. A habit tracker. A booking page for your salon. A directory of local dog parks. These are genuinely valuable products, and not one of them needs an AI brain to work. They might add a sprinkle of AI someday (“suggest items based on my past lists”), but that’s a feature you bolt on later, not a thing you need today.

So here’s the permission you came for: if your app doesn’t need a brain, don’t get one. Read the rest of this lesson so you understand the piece, then skip the signup and move on. Nothing later in the course breaks because you waited. You can always come back to this exact lesson the day you decide to add an AI feature.

For our trip planner, and for most of the apps people bring to this course, the brain is the magic. So let’s understand what it costs, and then go get one.

Remember the two AIs from the Quickstart, and how they cost money in separate ways? This is where that matters.

Your builder runs on your Claude subscription, one flat monthly price that you pay. Your app’s brain is different. It’s billed by whoever provides it, and it charges per use. Every time one of your users generates a trip, that’s one tiny charge against your account. Ten users, ten charges. A thousand users, a thousand charges. The bill grows with how much your app gets used, which is actually the good kind of bill: you only pay when people are getting value.

Two reasons this is a separate account and a separate bill. First, the brain runs inside your app for strangers, while your subscription is just for you, two completely different relationships. Second, and this trips people up: using Claude Code does not give your app free access to an AI brain. Even if you choose Anthropic’s Claude as your brain, that’s a separate signup with its own billing, unrelated to the Claude subscription that powers your builder. Two doors, two keys, two bills.

The good news for right now: it’s cheap to start, often free. The providers we recommend have generous free tiers that are more than enough to learn on, and once you’re paying, it’s a fraction of a cent per request. Later, in Lesson 21, we’ll set up guardrails so a runaway bill can never surprise you. For today, you’re looking at pennies, if anything.

To let your app talk to a brain, your app needs a key. Think of it like a membership card: it lets your app into the AI service and proves it’s allowed to be there.

We recommend Google Gemini, because it’s the cheapest and has the most generous free tier, which is perfect while you’re learning. But everything in this course works with any major AI provider, so if you already have a favorite, use it. Your app barely cares which brain it’s talking to.

One important rule, and we’ll come back to it in Lesson 3 when we cover what we call the secret leak: when you get this key, save it somewhere safe like a notes app or a password manager, and never paste it into your code. For now, just get the key and keep it handy. We’ll wire it up safely later.

You’ve made the call. Either your app needs a brain and you’ve got a key saved safely, or you’ve decided it doesn’t need one yet and you’ve saved yourself the signup. Both are the right answer, because both came from understanding what the brain actually does instead of just doing what a tutorial told you to.

Now comes the fun part. In Lesson 1, “How to Direct Your AI Builder,” you’ll build your first real, working app in a single conversation. You’ll learn the director mindset, how to describe what you want clearly enough that Claude Code builds exactly the thing in your head. If you got a key, you’ll hand it to your app and watch a real itinerary appear on screen. If you didn’t, you’ll follow along just as well and tell Claude Code to leave a placeholder where the brain would go.

It’s the most important skill in this whole course, and you’re about to start practicing it. Let’s go direct your builder.