Abstract: Large Language Models (LLMs) have recently shown promise in compiler optimizations such as loop vectorization and memory access restructuring. However, due to their generative nature, ...

Gary Tan reveals how to leverage the harness in order to achieve 10-100x productivity gains with the same AI model.

Write .gft files to define natural-language I/O pipelines — code review, ideation, content generation, data analysis, debate architectures. The compiler generates .claude/ harness structures — agents, ...