01, Context

The situation

Dragon Lords was launched by Mighty Dragon Studios in 2011 and exceeded one million downloads across iOS and Android. When the original publisher closed the game in 2019, a group of long-term players came together to bring it back and the source code was released to them. I led the development effort for that revival.

The live game back end comprised approximately 130,000 lines of PHP written for PHP 5.5. The code existed in fragmented and inconsistent versions, some source was missing, and the system had no established PHPUnit test environment.

02, Recovery

The work

I consolidated the code, reconstructed missing source from first-hand product knowledge and created a living technical specification. I introduced unit tests, database abstraction and dependency injection, and began moving procedural code toward typed, object-oriented PHP 8.3.

03, Conversion

The AI-assisted upgrade

Using AI-assisted engineering, I completed the initial PHP 5.6 to PHP 8.2 conversion in 14 days. The work included replacing thousands of inconsistently written calls to outdated database-access libraries with consistent modern equivalents.

04, Control

Human review and quality control

The conversion was divided into batches small enough for human sense checking. I checked the early batches heavily, spot-checked the remaining batches and applied an automated quality gate using a test suite developed beforehand with AI assistance.

Initial testing found few bugs, providing practical evidence that controlled batches, human review and automated tests had kept the conversion safe.

05, Outcome

The result

The recovered system was re-released and continues to be modernised toward typed PHP 8.3. Since taking over the game, I have added many new features, with overwhelmingly positive feedback from returning and new players. The work has established a safer basis for further feature development, performance investigation and live-player support.

06, Relevance

What this proves

AI can accelerate a difficult legacy-software investigation and upgrade when it is paired with domain knowledge, controlled batches, tests and accountable human review. The method is designed to recover understanding and reduce change risk, not to replace engineering judgement.