This is the moment the tool stops being the interesting part. The prototype works, someone with authority likes it, and the question becomes whether it can carry real users and real data.
Emergent scores 8.1 on our 361 score and is genuinely good at producing working applications. Whether a specific generated application is production-ready is a separate question, and it is answerable.
First: do you own it
Export the code to GitHub if you have not already. Emergent supports this and offers its own hosting as an option rather than an obligation — you can host anywhere.
Until the code is in a repository you control, everything below is theoretical. Ownership is the precondition for the rest of the checklist, and it is also the cheapest insurance you will buy.
Security — the non-negotiable list
- Secrets: confirm nothing is hardcoded. Environment variables, and rotate anything that was ever committed.
- Authentication: verify it is real, server-enforced authorisation rather than a UI that hides buttons.
- Input handling: check that user input reaching the database is parameterised.
- Dependencies: run an audit. Generated projects can pull in packages nobody chose deliberately.
Data — the part that is hard to undo
Backups first, and a restore you have actually tested. An untested backup is a belief, not a capability.
Then look at the schema honestly. Generated data models are usually reasonable for the feature that prompted them and less considered about what comes next. This is the cheapest moment to fix it — migrations get harder once real records exist.
If the application holds personal data, the compliance questions apply exactly as they would to code your team wrote. Provenance changes nothing about your obligations.
Observability — you cannot fix what you cannot see
Prototypes rarely ship with error tracking, structured logs or uptime monitoring, because nothing prompted them to. Production without these means your first notification of an outage is a user telling you.
Add error tracking, request logging and an uptime check before launch, not after the first incident. This is a short task that consistently gets deferred until it is expensive.
Hosting and deployment
Decide where it runs. Emergent's hosting is convenient; moving to something like Netlify (7.1) or your existing platform gives you the deployment pipeline, rollback and environment separation that production work assumes.
The specific requirement is boring and important: you need to be able to roll back. If the answer to "how do we undo a bad deploy" is unclear, you are not ready regardless of how good the application is.
The human handoff
Name the person who maintains this. Not a team in the abstract — a person who has read the code and can change it.
That review is the real gate. Code nobody has read is a liability the first time it breaks at an inconvenient hour, and the fact that an AI wrote it does not change who gets called.
Budget time for this. The most common failure with generated applications is not bad code; it is a working application with no owner.