Best-of-Pool: how we combine an LLM with a genetic algorithm to solve ARC-AGI-2 puzzles — the idea, a real solved task, and the data behind our paper.