RESCUING SOFTWARE BUILT WITH AI

From a quick prototype to robust software.

Software built with AI can be a valuable starting point. Whether it can be developed professionally depends on its code, architecture and operational requirements. We assess what holds up — and what is still missing.

THE STARTING POINT

Developed quickly. Not finished yet.

In what is known as vibe coding, software is largely generated through natural-language instructions to AI tools. This makes it easy to try out ideas quickly. An impressive demo alone does not show whether the resulting code will be maintainable in the long term.

Adding features one by one can lead to inconsistent patterns, duplicated logic or unclear responsibilities. Professional further development therefore starts with an assessment and a shared understanding of the intended product.

WHAT WE EXAMINE

Not just whether it works. But what it is built on.

  • Architecture and data model

    Are responsibilities clearly separated? How is data stored, changed and exchanged between components? We look for dependencies that make further changes unnecessarily difficult.

  • Access controls and security

    We review authentication, roles, server-side permissions checks and the handling of sensitive data. Hiding a button is no substitute for access control.

  • Tests and interfaces

    Critical workflows, error cases and external integrations are checked in a traceable way. Interfaces also need to handle timeouts, invalid responses and retries.

  • Production operation

    Configuration, updates, logging, backups and recovery are part of the assessment, along with the question of who will maintain the application.

THE PATH TO COMPLETION

Keep what works. Improve what needs attention.

We classify components by their potential for reuse and prioritise obstacles to production use. Working parts can remain in use where quality, rights and dependencies allow. Problematic parts can be replaced selectively.

The assessment produces a defined action plan. For a customer area, for example, separate roles, tested data access and a traceable ERP exchange could be agreed as acceptance criteria. The specific scope depends on your project.

Completion also means readiness for handover: source code, configuration, documentation and operational knowledge must be available and understandable to the people responsible.

THE GOAL

You know how your prototype can move forward.

Explore the separate, paid project assessment
  • An assessment of the existing architecture and code quality.
  • A distinction between reusable components and those that need replacing.
  • Prioritised actions for security, features and integrations.
  • A proposal for testing, documentation and verifiable acceptance.

FREQUENTLY ASKED QUESTIONS

What you
should know.

In principle, yes. The deciding factors are the existing code, its quality, usage rights, external dependencies and requirements. An assessment may recommend targeted further development, a partial rebuild or a fresh start.

No. Replacing everything is not the starting point of the assessment. We examine individual components and their relationships. Where a targeted fix would not be sensible or economical, we explain why replacement is needed.

Useful materials include source code with its change history, requirements, documentation, a test environment and an overview of data, services and responsibilities. Usage rights and permitted access need to be clarified. Confidential documents or credentials should not be entered in the initial contact form.

When the agreed features and quality criteria have demonstrably been met and handover to the intended operating environment has been clarified. A working demonstration alone is not enough. We define the specific acceptance criteria for each project.

THE NEXT STEP

Let’s assess where your prototype stands.

Discuss your project