🔗 Reading code with AI, not generating it

Instead of relying on AI to generate code, a back-end developer at Spatie utilized AI to understand an extensive legacy codebase. After joining a project with over 150,000 lines of code, the developer found that leveraging existing AI workflows was key. They used tools like Laravel Boost for real-time schema access and relied heavily on Claude to navigate and make sense of the structure and intricacies of the codebase.

Prompts tailored to understand the project’s architecture, core data models, and specific domains allowed the developer to efficiently onboard and contribute to the project. By embracing a domain-driven development approach, the new hire was able to work effectively within a well-defined structure, ultimately reducing the typical learning curve significantly.