FROM DEMO TO LIVE

Taking an AI-built app live: what needs to be in place?

You built something that works with Lovable, Bolt, Cursor or Claude Code, and now you want real users on it. This is what has to be in place before that moment, as a list you can tick off yourself.

The short answer

An app that works in the demo is not yet ready for real users. As long as you are the only one using it, that does not matter. With real users and real data, it does.

Before going live, five things have to be clear: who can reach which data, where the keys and passwords are, whether there is a way back, whether changes land somewhere else before they reach your users, and who notices when the app goes down.

Wondering how your own app holds up? Send me the link on WhatsApp. Within one working day I'll tell you, free of charge, whether a review makes sense.

What the demo does not show

With AI tools, something that works exists within days. That is real progress: the idea is proven and you can see what you want.

What the demo does not show is what sits underneath. Who can reach which data? What happens when a change breaks something else, and is there a way back? Those are not questions about the tool you built with, but about what was checked afterwards.

The go-live checklist

Go through these eight points before you share the link.

  1. Each user sees only their own data, even when they change a number in the address bar.
  2. Keys and passwords are not in the code and not in the browser.
  3. There is a test environment with fictitious data, separate from the real one.
  4. A backup has been restored once, so you know it works.
  5. Someone gets a message when the app goes down at night.
  6. The domain, the hosting and the accounts with external services are in your company's name.
  7. The parts the app leans on are up to date.
  8. You know whether the app processes personal data or payments, and where that data lives.

If you cannot tick a point, there is no reason to panic. It is a reason to sort it out before the first user signs in.

What stays and what gets replaced

What is good stays. I rebuild only when there is a clear reason.

First I capture in tests what the most important parts do today. Then it is safe to tidy up: split files that have grown too large, make database changes repeatable, separate test from real.

At the end there is one place with the code that another developer can open, start and understand.

Keep building yourself after launch

You do not have to hand over the building. Many owners want to continue themselves with Claude Code or a similar tool, and that works.

In that case I set up the route changes take to go live: automatic checks on every change, a test environment with fictitious data, and only then a controlled step to the real app. AI tools get no access to the real database. I enforce that technically, not just with an instruction.

MyFIFOGuide works this way: the owner makes changes with Claude Code and publishes them himself.

What a baseline review is

A baseline review is these points checked by someone who reads the code. You get a short list of findings in three kinds: what has to happen first, what is important and what can wait.

The baseline review costs €495, excluding VAT, for one app. You get the findings within five working days, with a 30-minute call in which I walk you through them. If you continue with me within 30 days, the full €495 comes off the price.

What I need: read access to your code and two test accounts. No passwords, no keys and no customer data. Afterwards you remove my access.

What I do not promise

No review guarantees that software is free of errors or fully secure. A baseline is also not a certification such as ISO 27001.

What you do get: you know what is there, which risks exist and how a change goes live in a controlled way. If you process sensitive data, an independent penetration test may be needed on top.

Want to know whether your app is ready?

Send the link to your app. Within one working day you hear, free of charge, whether a baseline review makes sense, followed by a short call and a fixed price.

Send a WhatsAppContact form

A link, a screenshot or a voice message is enough. Usually a reply within one business day.

If an idea is unnecessarily complicated or expensive, I say so before I start building.