AI Systems Collaborate to Build SQLite-Like Database Engine in Rust

Summary

A developer describes an experiment where they tasked three AI systems (Claude, Codex, and Gemini) to collaboratively build a SQLite-like database engine in Rust. The resulting codebase is approximately 19,000 lines and implements core database components including a parser, planner, volcano executor, pager, B+ trees, write-ahead logging, recovery, joins, aggregates, indexing, transaction semantics, and statistics-aware planning. The project includes 282 AI-written unit tests and supports a subset of SQL, though it likely fails against the official SQLite test suite. The article appears to be a response to a Hacker News discussion about whether such AI-generated code actually works.