Prototype review
You built an app with an AI coding tool and it works. Find out what it needs before other people rely on it.
AI coding tools let the person who understands a problem build a working answer to it in days. That is worth a great deal, because the hard part, knowing what the software should do, is already proven.
What a prototype does not tell you is whether it is safe to put in front of a team, connect to real data or leave running unattended. A prototype review answers that. We read the code, run the app and tell you in plain language what to keep, what to fix and what to rebuild, with a fixed scope for the work if you want it done.
What we look at
Nine areas, the same ones every time, so nothing is missed because it happened to look fine. The reasoning behind each is in our article on taking a vibe-coded prototype to production.
Sign in and access
Who can get in, what each person can see and change, and whether those rules are enforced on the server or only on the screen.
Secrets and settings
Passwords and keys kept out of the code, with separate settings for testing and live use.
Data
Where it is stored, how it is structured and checked, and whether a backup has ever been restored.
Code and dependencies
Known weaknesses in the code and in the libraries it pulls in, and the licences those libraries carry.
Tests
Whether anything confirms the app still works after a change, especially on the paths people depend on.
Failure and recovery
What a user sees when something breaks, what gets recorded and who finds out.
Deployment
How a new version reaches users, whether someone else could repeat it and how it is rolled back.
Privacy and records
Any personal or commercially sensitive information, where it is held and whether a change can be traced to a person.
AI features
If the app itself calls an AI model, how it copes with misleading input, wrong answers and runaway cost.
What you get
- A written report in plain language, short enough to read in one sitting.
- A clear call for each area. Keep it, fix it or rebuild it, with the reason.
- The risks ranked, so the few that matter are not buried among the many that do not.
- A fixed scope for the work to make it production ready, which you can take up with us, with your own team or with someone else.
- A conversation to walk through the findings and answer questions.
How it runs
Share
Access to the code and the running app, under a confidentiality agreement if you want one.
Read and run
We read the code, run the app and test the paths people depend on.
Report
Findings in plain language, area by area, with the risks ranked.
Decide
You choose what happens next. There is no obligation to build with us.
Share
Access to the code and the running app, under a confidentiality agreement if you want one.
Read and run
We read the code, run the app and test the paths people depend on.
Report
Findings in plain language, area by area, with the risks ranked.
Decide
You choose what happens next. There is no obligation to build with us.
Who it suits
- You built a tool for your own job and now the whole team wants it.
- Your business paid for a prototype and needs to know what it will take to go live.
- An internal app has grown past the person who made it, and nobody is sure what is holding it up.
If you want the work done, making a prototype production ready is one of the four ways we work, set out on our Software and AI page.
Common questions
Will you tell us to throw it away and start again?
Rarely. The screens and the workflow are usually the valuable part, because real use has already tested them. What tends to need rebuilding sits underneath, such as how data is stored and how access is controlled. The report says which is which and why.
Can we keep using AI tools to develop it afterwards?
Yes, and we would expect you to. Part of the work is leaving the app in a state where AI tools produce safe changes, with tests that catch a mistake and settings kept out of the code.
What do you need from us?
Access to the code, a way to run the app or a link to it, and half an hour with the person who built it. A list of what users have asked for helps too.
Is our code kept confidential?
Yes. We work under a confidentiality agreement where you want one, and the code and the report stay yours. Where we use AI tools to help read it, that is under terms that keep your code out of model training.
What does it cost?
The review is a fixed scope, priced before we start and sized to the app. Tell us what you have built and we will give you a figure.
Request a prototype review
Tell us what you built and we will come back with a scope and a price.