Building your own app with AI.
The tools, the connectors, the models.

What to download, and why each exists.

1. A place to write and run code.

   Claude Code runs in your terminal and can read, write, and run your project. Cursor and VS Code are editors that an AI can work inside. Why: the AI needs a place where it can act on files, not just talk about them.

2. The platform's own kit.

   Xcode for iPhone and iPad (Mac only), Android Studio for Android. Both include the simulators you test on. Why: the stores require builds made with the official tools, and the simulator is how you see what a reviewer will see.

3. A cross-platform layer, if you want both stores from one codebase.

   Expo and React Native let one project ship to both. Why: two native apps means two of everything; one shared codebase means one.

4. A backend.

   Somewhere for accounts, data, and payments to live: a hosted database with authentication and a payment provider. Why: the app on the phone is the front of the house; the backend is where the business actually is.

5. Version control.

   GitHub, through the desktop app or the terminal. Why: every change is saved with a history, an AI can propose changes you review, and nothing is ever lost.

How an AI reaches your tools.

A connector (the technical name is an MCP server) is a small bridge that lets an AI tool talk to a specific service: your repository, your database, your payment account, your browser. Without connectors, the AI can only write text and you carry it everywhere. With them, it can open a pull request, read a table, check a payment, or load a page. Why they matter: the difference between an assistant that advises and one that does.

Claude or ChatGPT guiding your computer.

There are three levels. First, the AI writes and you paste. Second, the AI works inside your project through a tool like Claude Code and asks before it runs anything that matters. Third, the AI drives your browser or desktop through a connector, clicking and typing where you would. Each level trades speed for oversight. Why the levels exist: the second is where most real building happens, because the AI can build and verify while you keep the deploy button.

• The AI proposes; you approve anything that spends money, touches customer data, or ships.
• Everything it does lands in version control so you can see it and undo it.
• It verifies by running, opening, or reading back, and says which.

Plan with the strongest. Build with the efficient one.

Use the strongest model available to you for planning, architecture, and review, because those decisions are expensive to get wrong. Use a fast, efficient model for the slices of work the plan produces, because that is where the volume is. Use the cheapest for repetitive checks. Why: cost tracks volume; judgment tracks stakes. Matching the model to the stakes is most of the cost control in a build.

Claude's models are built to work inside your project through Claude Code; ChatGPT is strong for drafting, explaining, and working through a problem in conversation. Many builders use both: one to think with, one to build with. Which is right depends on the tool you already live in.

The boring layer, and when to hire it.

A prototype that runs and a product that survives are different things. The difference is the boring layer: support, testing, monitoring, backups, access control, a data model that will not fall over in month three, and the store approvals. Building a simple app yourself is realistic. Building a platform your business depends on, with payments, integrations, and a review to pass, is a project, and it is what we do at Front Page Intelligence. The guides here teach the map. They do not hand you a copy of our systems, and they will not turn a weekend into a business platform. They will make you a better client, or a capable builder of the right-sized thing.

Describe the idea.

Four questions before any tool gets installed. Entries stay on this page and are not sent anywhere.

A plain-language build plan for your idea.

Paste the four answers. It tells you what you need, what it will cost per month, what to build first, and whether this is a weekend build or a project to hire.

I want to build an app with AI. From my answers, give me a plain-language plan. Do not write code and do not give step-by-step setup instructions. Tell me what each thing is and why I need it.

1. THE TOOLS: which of the five kinds I need for my case (a place to write and run code, the platform kit, a cross-platform layer, a backend, version control), named, with one line on why each.
2. THE CONNECTORS: which services my AI tool should be connected to, and what each door opens.
3. THE MODELS: which kind of model plans, which builds, which checks, and roughly what a month of building costs at my size.
4. BUILD ORDER: the first three things to build, in order, and the first thing to show a real user.
5. THE HONEST CALL: weekend build, a few weeks of evenings, or a project to hire, and why. Name the boring layer items I will need before this can hold real customers.

MY ANSWERS:
[what it does and for whom; where money and data live; which stores; my machine and comfort with a terminal]

One prompt. Paste it into Claude. It builds the whole thing for your business.

The Pro pack is a build prompt with blanks for your context: your product, your medium, your team. Fill them in, answer its questions, and it produces the complete system in your words. The PDF explains the method so you can judge the output. The setup order for Mac and Windows, the connector map, the model routing table with costs, the let-it-drive rules, the project brief, and the ship checklist.