With Cursor, Claude Code, Lovable or Bolt, you can go from idea to a working application in a weekend. Logins work, data saves and the dashboard looks great. It's tempting to conclude the product is ready to launch.
It may well be. But "it works" and "it's production-ready" answer two different questions.
"Working" means the software does what it should when everything goes right. "Production-ready" means it keeps doing it, safely, when things go wrong.
What "working" actually proves
A working demo proves the happy path: a user signs up, does the thing and sees the result. It's tested by the person who built it, with a handful of records, on a good day.
It says nothing about what happens when:
- a user changes an ID in the URL and sees someone else's data,
- the database is accidentally deleted at 2am,
- a dependency publishes a vulnerable update,
- traffic grows tenfold after a mention on LinkedIn,
- an AI API call loops and runs up a five-figure bill overnight,
- the one person who understands the code is on leave.
The six questions production readiness asks
At Bronkode, we break production readiness into six pillars. Each one asks a question a demo can't answer:
- Security: can the wrong person get in, or the right data get out?
- Architecture: does the design hold up as usage and features grow?
- Reliability: when something breaks, how quickly and completely do you recover?
- Engineering maturity: can changes be made safely and repeatably?
- Operational readiness: will you know something is wrong before customers tell you?
- Cost & vendor risk: will success bankrupt you, or a provider hold you hostage?
Why AI-built software makes the gap wider
AI coding tools are optimised to satisfy the prompt in front of them. Ask for a login page and you get one. You don't automatically get rate limiting, account lockout, audit logging, a backup policy or a deployment pipeline, because nobody asked.
Experienced engineers carry that invisible checklist in their heads. When software is generated rather than engineered, the checklist is often missing, and nothing on screen tells you so.
What to do about it
You don't need to become an infrastructure expert. You need an honest picture of where you stand:
- Run through the AI Application Production Readiness Checklist yourself.
- Fix the critical items: secrets, access control and backups first.
- Get an independent review before you take payments or sensitive data.
Working software is a great achievement. Production-ready software is what lets you keep it.