A 10-point readiness check

Is your AI-built app ready for customers? Ten checks you can do yourself.

"It works" and "it's ready" are different bars. These ten checks cover ownership, security, reliability, performance and deployment, the same five areas I use in a readiness review. You don't need to be technical to start.

FIXED PRICE  ·  FOUNDER-LED  ·  READ-ONLY ACCESS, NDA IF YOU WANT ONE

Ownership

Do you actually control it?

1. Can you log in to everything?

Open the code repository, hosting, database and domain registrar. If any of them sit in a freelancer's or an AI tool's account rather than yours, that's a finding.

2. Could a new developer pick it up?

Is there a README, and could someone run the project from it in a day? If nobody on the team can explain how login or payments work, that's one to flag.

Security

Who can see what?

3. Can one customer see another's data?

Create two test accounts. Log in as the first, find a page with an ID in the address, and swap in an ID that belongs to the second. If it loads, stop and fix that first. On Supabase, start with the security checklist.

4. Are secrets sitting in the code?

Search the repo for API keys, passwords and .env files that were committed. Open your live site's developer tools and look through the loaded scripts for anything that looks like a key.

5. Would you notice a failed payment?

Make a test payment that fails. Does anything log it, and would you see it? Failed charges that nobody sees are the ones customers write angry emails about.

Reliability

What happens when it breaks?

6. Do you have backups, and have you restored one?

A backup you've never restored is a hope. Find out what's backed up, how often, and who could bring it back.

7. Would you know it's down before a customer told you?

Is there an uptime monitor or an error alert that reaches you? If your first warning is an email from a customer, the answer is no.

Performance

What gives out first?

8. What happens when usage doubles?

Which screen gets slower as the data grows? A demo with ten rows hides problems that a hundred thousand rows will find. Ask the person who built it where they'd expect it to creak.

Deployment

Can you change it safely?

9. Is there somewhere to test before customers do?

A staging copy of the app that isn't live. Without one, every release goes straight to the people paying you.

10. Can you undo a bad release?

If today's change breaks something, can you get back to yesterday's version in minutes? If the honest answer is 'I'd have to ask someone', that's a blocker.

Your score

What the answers mean

Most AI-built products answer "not sure" to several of these, because the tool never asked. That's fine. What matters is which ones block your next milestone, whether that's your first paying customers, an enterprise pilot, investor due diligence or a launch date.

Fix what blocks the next milestone. Leave the rest alone for now.

If checks 3, 4 or 10 worry you, start there. If you'd rather someone else did the whole list and told you in plain English what to do about it, that's the AI Product Readiness Review: £1,950 fixed, five working days from access, and useful even if you hire someone else to do the work.

Common questions

Readiness checks, answered

Do I need to pass all ten checks before launching?

No. What matters depends on your next milestone. A handful of blockers, like customers seeing each other's data, need fixing first. Plenty of the rest can wait. Separating the two is what the NOW vs LATER roadmap is for.

I'm not technical. Can I still do this?

You can do more than you'd think. Several checks are just logging in with two accounts or opening a settings page. For the rest, 'I don't know' is a useful answer, and it's the part a review covers.

What if I answer 'not sure' to most of them?

That's common with AI-built products, because the tool never asked you the questions. It isn't a failure. It means a review will be worth more than a rebuild.

Does it matter which AI tool built my app?

Not much. The checks are the same whether it came from Lovable, Bolt, Replit, v0 or Cursor, because they test the product, not the tool.

Keep reading

Related guides

How to fix a vibe-coded app

What breaks first, and when not to rebuild.

Read the guide

Supabase security checklist

Ten minutes to check Row Level Security.

Read the checklist

AI Product Readiness Review

A fixed-price review against your next milestone.

See what's included
Ready when you are

What's your next milestone?

Tell me what's changed commercially, and when. If a review makes sense I'll say so. If it doesn't, I'll say that too.

Darryll Wolloff is the founder of TechConnex. He builds MVPs and fixes AI-built apps for early-stage founders, and still writes and ships production code himself. Last updated 17 November 2026.