MindPalaceX

Free · no card · the whole thing

Get yourself set up to build with AI.

This is the exact setup I'd do with you, written out. Every account, the order to open them in, what each one is for, and the mistakes that kill a first build.

Nothing is held back and there's nothing to buy. Read it, do it, keep it.

Want it in your inbox as you go?

I'll send this as a checklist plus a short series that walks through it with you — one email per step. You can also just read the page. It's all here.

Six tools, and what each one is actually for

None of these names need to mean anything to you yet. Each one does exactly one job.

Think Claude Where you work out what you are building before you build it.
Work Cowork Where Claude gets access to your actual tools and information.
Build Claude Code Where the software gets written. You describe, it builds, you review.
Save GitHub Where your code lives, so a bad afternoon is never permanent.
Data Supabase Where your app remembers things — users, records, everything.
Publish Netlify Where it becomes a real address other people can open.

Claude, Cowork and Claude Code are one account between them — that's why the list below has four sign-ups and this list has six tools.

The setup, in order

Do them in this order. The order is most of the value.

  1. 1

    Open the four accounts

    about 10 minutes

    Do this before anything else and in one sitting. Signing up halfway through a build is how an evening disappears.

  2. 2

    Install Claude Code

    about 10 minutes

    Claude Code is the part that actually writes software, and it runs on your own machine. Install it, sign in with the same Claude account, and open it in an empty folder. If a terminal is new to you, this is the step to slow down on — everything after it is typing in plain English.

  3. 3

    Write down what you are building, before you build it

    about 30 minutes

    One user, one painful problem, one result. Not a feature list. If you cannot say "it worked if, 30 days from now, ___" in a single sentence, you are not ready to build yet and no tool will save you. The free Workshop below turns a rough idea into a proper spec in about five minutes if you would rather not start from a blank page.

  4. 4

    Connect the database and the deployment first

    about 20 minutes

    Before a single feature. Create the Supabase project, connect the Netlify site, get an empty page live at a real URL. Almost everyone does this last and it is the single biggest reason first builds die: a project that has never been deployed is a project where deployment is an unknown risk sitting at the end.

  5. 5

    Build the smallest version that a real person could use

    an afternoon

    Not the version you want. The version where one person can do one useful thing end to end. Paste your spec into Claude Code and ask it to build only the core loop — sign in, do the thing, see the result. Everything else is v2, deliberately.

  6. 6

    Put it in front of one real person

    this week

    Not five. One. Watch them use it without helping. What they get stuck on is your next week of work, and it is never what you expected. This step is free, takes an hour, and is the one people skip for months.

Running cost once you're set up: about $20 USD a month. Claude Pro is the only paid piece and it covers Claude, Cowork and Claude Code together. GitHub, Supabase and Netlify all have free tiers a first build sits inside comfortably.

The first prompt

Paste this into Claude Code once your spec is written. It stops the most expensive mistake — building for an hour before anyone has agreed what "smallest useful version" means.

I am building a small web application. Here is what it does: [paste your spec here] Before writing any code: 1. Tell me the smallest version of this that one real person could use end to end. 2. Tell me what you are deliberately leaving out of that version, and why. 3. Ask me anything you need to know that I have not told you. Do not start building until I have answered.

The five things that kill a first build

I have watched every one of these happen more than once.

Building the version you want instead of the version that works The first one exists to be used by one person. Everything you add before that is a guess with no feedback attached.
Leaving deployment until the end Get an empty page live on day one. The risk you defer is the risk that kills the project.
Not saving your work to GitHub It takes ten minutes to set up and it is the difference between a bad afternoon and starting over.
Asking for everything in one prompt Ask for one piece, look at it, then ask for the next. AI is fast enough that small steps cost you nothing and save you rewrites.
Not knowing what "done" means Write the acceptance criteria before you build. Otherwise you will keep going forever and never ship.

Two more free things

Both take minutes and neither wants a card.

The Workshop → Turns a rough idea into a proper build spec in about five minutes. That's step 3 above, done for you, and the spec works with any AI builder — not just mine.
The diagnosis → Ten minutes of questions about your business, and it names the one system worth building first. That's the part of step 3 people get wrong most often.

And if you'd rather not do it alone

Plenty of people will read this page and get there on their own. That's a good outcome and it's why the page exists.

But two things on this page are genuinely hard to do from a document: choosing which system is the right first one for your business, and having someone there when step 4 doesn't behave the way step 4 is written.

That's what I sell. 2 to 4 working sessions where we find the gaps, build your foundation, and get one real system of yours running. Same map, with me walking it with you.

See what that costs →

$5,000 paid in full, or 2 × $2,750.

© 2026 David Owasi · OutreachGenius · Terms