This guide sells you nothing until step five. It shows you what to ask the AI for, what you'll get back, and the exact point where everyone gets stuck. Then we talk about where you leave it running.
Gather these first. People who arrive with all four settled finish the store in an afternoon; people who don't spend the afternoon deciding.
One sentence, written down. "Single-origin coffee from Antioquia for offices in Medellín" works. "Quality products" is useless.
Yours, not stock. It's the single biggest factor in whether people buy, and it's the one thing no tool will solve for you.
With VAT decided and shipping thought through. If it isn't clear now, you'll be redoing it later.
The name people will find you by. Buy it first: changing it later drags email, invoices and search rankings along with it.
This is the step that decides everything else. You don't ask an AI to "build me a store": you brief it the way you'd brief someone who is actually going to do the work. The more specific the brief, the fewer rounds afterwards.
Fill in these three fields and the brief writes itself. Copy it and paste it into Claude or whichever tool you use.
Build your brief
That last line is the important one. Asking it to explain what it built is what separates someone who ends up with a store they understand from someone who ends up with a folder they're afraid to touch.
What you get back will look finished. It isn't. Open it on your own phone and buy something yourself, start to finish, as if you knew nothing about the store.
Write down what breaks and hand it back to the AI in plain sentences: "the add-to-cart button is invisible on mobile", "the price doesn't show thousands separators", "after I pay it doesn't tell me what happens next". Three or four rounds of this and the store changes level.
Test it on someone who isn't you. You already know where everything is. Sit next to someone who doesn't, and stay quiet while they buy. That's where the real problems surface.
This is the boring part, and it's the one nobody will do for you. Your photos, your real prices, descriptions that say something. The store the AI handed you comes with sample products: leave them in and the store looks like a sample.
If there are a lot of products, put them in a spreadsheet — name, description, price, photo — and ask the AI to load them from there. It's faster than doing them one at a time, and you're left with a catalogue in a file you can reuse.
This is where almost everyone gets stuck, and not for technical reasons. Taking money online in Colombia requires being formally registered: a RUT (the Colombian tax ID), a bank account in the business's name, and a contract with a payment gateway. That takes days, not minutes, and it doesn't depend on the AI or on us.
Start that paperwork the same day you start the store, in parallel. Most people discover this step once everything else is ready, and that's where they lose the week.
You can sell in the meantime. Launch with bank transfer and confirmation over WhatsApp. It isn't elegant, but it takes money from day one and tells you whether the product sells before you've set up the gateway.
The store exists on your computer. For it to exist for everyone else it has to run on a server, all day, every day. There are two roads from here.
One is a platform that charges per user or per sale, in dollars, on a personal card. It works — and the day the dollar climbs or the store grows, the invoice climbs with it.
The other is a server of your own. We hand it over with full access, point the domain at it and leave it running. Fixed price in pesos, backups included, support in Spanish and English. If tomorrow you want to run the CRM next to it, it fits on the same machine.
Said plainly: this is the step where we come in, and it's the only one of the six where we sell you anything. The other five you can do without LaPoint.
Publishing isn't finishing. A store that sells is a store that has to be up at eleven on a Saturday night, that keeps your customers' orders safe, and that can be rolled back to how it was if a change breaks it.
Three things, and that's it: automatic backups, so you can go back. A security certificate, so the browser doesn't scare the customer off. And knowing who to call when something fails, in your language and in your time zone.
Saved in this browser, on your machine. We don't see it.
Everyone is telling you what it will. This is the other half, and it's the half that decides whether your store sells or merely exists.
Half an hour, no cost. If the problem is step five, we'll show you the server. If it's step four, we'll tell you which registration you're missing. And if the store you built works fine as it is, we'll tell you that too.
We set up the server, point the domain and leave it running.
We review what you're paying today and what it would cost in pesos. Nothing moves.
CRM, inventory, automations. They fit on the same machine.