Audit and fixing of AI-built projects

Audit and rescue of projects built with AI

If your app, website or automation was built with ChatGPT, Claude, Copilot or “vibe coding” tools and now fails, is insecure or nobody knows how to change it, we review it. We run a technical audit with a prioritised plan, fix what is critical, put it into production with tests and backups, or tell you honestly when it is better to rebuild it.

By Oliver Guirado · Updated on November 2, 2026 · 5 min read

Outline of rescuing a project built with AI: technical audit, a decision between fixing and rebuilding, stabilisation, going into production with a safety net and continuous improvement

Who the rescue of AI-built projects is for

It is for founders, small businesses and teams that built an app, a website, an automation or an agent with AI and have reached the limit of what the tool can solve: it fails with real users, every change breaks something else, security or data is a worry, or the supplier who built it is gone.

It is also hired by companies that have inherited AI-generated code and want to know, before investing more, whether it is worth carrying on with. The warning signs and the market data are in the guide fixing code and apps built with AI.

What we do

Technical audit

Review of the code, architecture, dependencies, data and performance, with a risk map in priority order.

Security and personal data

Exposed keys, permissions, validation of user data, backups and a GDPR review.

GDPR and the AI Act →

Stabilisation

We fix critical failures and add automated tests to the flows that cannot fail: sign-up, payments, data submission.

Refactoring and migration

We tidy the structure, update dependencies or migrate to a solid base if the code cannot be saved.

Custom development →

Production with a safety net

Separate environments, repeatable deployment, monitoring, alerts and a recovery plan.

Rules for carrying on with AI

A guide for your team: what can be asked of AI, what a person always reviews and which tests are mandatory.

Implementing AI with control →

How we work

  1. First call and confidentiality. You tell us what the project does, what fails and what worries you. If you need it, we sign a confidentiality agreement.
  2. Read access and audit. We review the code and the environment without touching anything, in one to two weeks depending on size.
  3. Report and decision. You receive the prioritised risks, the real state and an honest recommendation, with the cost and timeline of each option.
  4. Fixing. We carry out the agreed plan, with partial deliveries and review with you.
  5. Handover. Tidy code, tests, minimal documentation and, if you want, ongoing maintenance.

What you receive

  • Audit report with the security, data, performance and maintenance risks, in order of severity.
  • Prioritised plan with the options (fix, stabilise or rebuild), their cost and their timeline.
  • Fixed code with automated tests on the critical flows.
  • Minimal documentation so someone else can continue the project.
  • AI usage guide for your team, with review and quality rules.

Sample project: rescuing an AI-built project

Sample project · Start-up

Stabilising an AI-built MVP for a booking start-up

Problem
The application had been built in a few weeks with AI tools and worked well in the demo. With the first customers came double charges, crashes at peak use, keys visible in the code and no backup.
Solution
A two-week audit and a stabilisation plan: fix the payment flow, move the keys to a secure store, add tests to the booking and payment flows, set up a test environment and switch on backups and alerts.
Steps
  1. Audit of the code, the database and the deployment.
  2. Report with prioritised risks and a proposed scope.
  3. Payment and security fixes, with automated tests.
  4. Separate environments, backups and monitoring.
  5. Handover with documentation and an AI usage guide for the team.
Target outcome
A stable product in production, with no known critical risks and a base on which to keep adding features with judgement.
Technology
Code review, automated tests, secrets management, test and production environments, monitoring and backups.

The projects on this page are illustrative, anonymised examples of how we approach this kind of work. Every real project is defined with your information, your systems and your goals.

Reference prices for rescuing AI-built projects

ServiceIndicative investmentTypical timeline
Technical audit of code, security and architecture, with a report and plan€1,500 – €6,0001 – 2 weeks
Stabilising an AI-built MVP: critical failures, tests and a safe move into production€5,000 – €25,0003 – 8 weeks
Partial rebuild or migration to a solid base€15,000 – €60,0002 – 5 months

Indicative reference prices in euros, excluding VAT, for projects in Spain and Europe in 2026. The final amount depends on scope, integrations and data quality; we confirm it with a fixed quote after a first meeting.

If after the audit you decide to continue with us, we deduct its cost from the fixing project. To start a new AI project on a solid base, see custom AI app development or request a quote.

Access, confidentiality and tools

We start with read-only access and sign a confidentiality agreement if you need one. We do not send your code to external AI tools without your express permission. When we use AI to work faster, a person reviews and tests every change before delivering it, which is exactly what we ask for your project. More in AI integration into applications and in AI consulting.

Frequently asked questions

What is rescuing an AI-built project?

It is the technical review and fixing of an app, website or automation built with AI tools that fails, is insecure or cannot be maintained. It includes audit, stabilisation and, if needed, a partial rebuild.

Can an app built with ChatGPT, Claude or vibe coding tools be fixed?

Usually yes. The audit determines whether it is enough to stabilise it or whether the critical parts should be rebuilt. With a prototype that has no users, rebuilding usually works out better; with an MVP customers already use, stabilising is the usual route.

How much does it cost to fix an AI-built project?

An audit costs between €1,500 and €6,000; stabilising an MVP, between €5,000 and €25,000; and a partial rebuild or migration, between €15,000 and €60,000. These are reference prices excluding VAT, confirmed after a first review.

How long does it take?

The audit takes one to two weeks. Stabilisation takes between three and eight weeks depending on the size of the project and the state of the code.

What do you need to get started?

Read access to the code or repository, a description of what the project should do, where it is deployed and what problems you see. If there is personal data, we first agree how to handle it.

Do you use AI to fix the code?

We use it as a support tool, but every change is reviewed and tested by a person before delivery. Your code is not sent to external AI services without your permission.

Shall we talk about your project?

Tell us about your case and we will reply within 24-48 hours with a proposed scope and an indicative quote.

Request a quote

Tell us about your project

Describe what you want to achieve and which systems you work with. We reply within 24-48 hours.

Fields marked with * are required

Artificial intelligence services and guides

Everything we do with AI for business, from the diagnostic to the product in production. Artificial intelligence for business.

Services

Guides

See all guides →

AI by sector

See all industries →

Technologies