We Programmers Need "Results" (rockyj-blogs.web.app)

A developer reflects on how AI (Claude, etc.) has made finding answers instant but argues we still need disciplined engineering: clear code organization, small stateless building blocks, and a reliable way to compose them. He recommends separating “foundational” interface code (controllers, DB entities/repositories, HTTP/Kafka/SQS clients, config) from service-level business logic where each function does one simple thing. This separation improves testability and prevents AI-generated code from becoming unreadable and unmaintainable. To safely tie these pieces together, he promotes using a Result wrapper (a Result monad) to handle nulls and exceptions uniformly. Instead of scattered ifs, try/catch and mutable state, you chain service calls with map/flatMap-style operations: errors and nulls are propagated automatically, code becomes declarative, smaller, and less bug-prone. The pattern is language-agnostic (TypeScript, Kotlin, etc.), keeps functions small and stateless, and makes composition consistent across projects—especially useful as more code is produced or suggested by AI tools. Full example code is on GitHub.