For founders shipping AI-built applications
Your AI-built app may look finished. Find out whether it is safe to launch.
An evidence-backed release audit for applications built with Claude Code, Cursor, Lovable, Replit, Bolt and similar tools. Get a clear launch verdict, the blockers that matter, and a prioritised fix plan.
Human-reviewed. Evidence-backed. Confidential by design.
A working interface is not proof that an application is ready
AI coding tools are genuinely good at producing something that runs. What they do not do is tell you which of the invisible parts were never finished. A screen that renders correctly proves the screen renders correctly. It does not prove any of the following.
Authorisation is enforced
That the check happens on the server, not only in the interface that hides the button.
Users cannot read each other's data
That every query is scoped to the right account, on every path, not just the ones you clicked.
Payments can be trusted
That your payment webhook verifies events actually came from your provider before it marks anything paid.
Deployments are recoverable
That the app can be rebuilt and restored, and that you would know how.
Failures are handled
That errors surface somewhere you will see them, rather than disappearing silently.
Tests mean something
That the passing test count reflects tests which actually exercise the critical paths.
None of this means AI-built software is bad. It means the gap between it works and it is safe to put in front of paying customers is real, and it is hard to see from the inside.
What you get
- Included:A go / no-go launch verdict
- Included:A 100-point readiness score across nine weighted categories
- Included:Verified launch blockers, each with the evidence behind it
- Included:Security and data-isolation findings
- Included:Build, test and deployment evidence — including what could not be checked
- Included:A prioritised remediation plan with effort bands
- Included:A fix-versus-refactor-versus-rebuild recommendation
- Included:A confidential PDF report you keep
How it works
- Step 1
Purchase and complete intake
Tell us about the application and choose how your code may be handled.
- Step 2
Provide read-only access
A read-only repository invitation, or an encrypted archive upload.
- Step 3
We inspect and run safe checks
Static review always; sandboxed execution only if you permit it.
- Step 4
A human reviews the evidence
Every critical and high finding is validated by the operator before it reaches you.
- Step 5
You receive the report
Verdict, blockers, fix plan, and a clear next step — including doing nothing.
The delivery clock starts when we accept the project and confirm working access — not at the moment you pay. If a project turns out to be out of scope, we say so and refund rather than deliver something thin.
Founding Release Check
£149First 5 projectsA tightly scoped release check for one web application: a go/no-go verdict, the blockers that matter, and a prioritised fix plan.
Target: within 24 hours of acceptance
Who this is for
- Included:A founder preparing to launch an AI-built application
- Included:A team inheriting AI-generated code they did not write
- Included:An agency validating a delivery before handover
- Included:An investor or buyer needing technical clarity
- Included:A product stuck between prototype and production
Who this is not for
- Not included:Anyone needing a formal penetration test
- Not included:Anyone needing regulatory certification
- Not included:Large enterprise estates
- Not included:Hardware, embedded or safety-critical systems
- Not included:Active malware analysis
- Not included:A guaranteed same-day repair
- Not included:Any repository you are not authorised to share