Guide
How to take an AI-built prototype to production.
Updated
AI tools can get you to a working demo in days. Getting from that demo to something real customers can sign up for, pay for and rely on is a different job. This guide lists what separates the two, so you can check your own prototype before you put real users or real payments through it.
Signs your prototype is not ready yet
- You are nervous about putting real users or real payments through it.
- Your web app works, and you now need it in the App Store and on Google Play.
- An investor, customer or security review has asked questions about the code that you cannot answer.
- Nobody is sure what happens when 1,000 people use it at once.
If one of those sounds like your situation, work through the checklist below.
The launch-readiness checklist
The demo works. Logins, payments, security and the app stores are another matter. Each item below is a place where a working demo and a launchable product are different jobs. Ask the question, and answer it honestly.
- Sign-in and permissions. Is authentication secure, with roles, password recovery and proper session handling?
- Data. Is the database structure sound? Are there migrations and backups, and has a restore been tested?
- Security. Are secrets out of the code, inputs validated and access rules checked? Have the dependencies been reviewed?
- Payments. Do subscriptions and one-off payments handle failures, refunds and receipts properly?
- Automated tests. Is there coverage on the paths that make you money, so that changes stop breaking things?
- Monitoring. Do error tracking, logging and alerts tell you about problems before your users do?
- Infrastructure. Are deployments repeatable? Are staging and production separate, with room to scale?
- Mobile release. If you need it: native iOS and Android builds, App Store and Google Play submission, offline behaviour and testing on real devices.
- Handover. Is the architecture documented, so your own team can run it?
How to score your own prototype
Go through the nine items and mark each one as sound, doubtful or missing. If you cannot show that something works, mark it doubtful.
Doubtful and missing items are what stands between you and launch, so put them in priority order: the ones that put users, money or data at risk come first. You do not need to score everything perfectly. A question you cannot answer is itself an answer, and it is the one an investor, a customer or a security reviewer will ask.
What to keep and what to rebuild
Your product decisions, screens and flows carry forward. What gets rebuilt is only the parts that cannot be made safe, and the checklist above is how you find them.
The AI tool that produced the prototype does not change any of this. What matters is the code and the running product, whatever produced them.
How to find out where you stand
A production readiness audit answers the checklist for your product. It reads your code, runs your product and tests it the way real users would. You get a written report that covers:
- What is sound and stays as it is.
- What has to be fixed or rebuilt before launch, in priority order.
- A fixed price and timeline to do that work.
Ours is the production readiness audit. $2,500 fixed. 3 to 5 days. The report is yours to keep and act on with anyone, and if you go ahead with us the $2,500 comes off the price.