Cooking a complex dish requires a clear recipe: first chop, then sauté, add spices at just the right moment. Chemical transformations also follow step-by-step recipes — mechanisms where atoms of carbon and hydrogen rearrange in a strict order.

Previously, neural networks often invented nonexistent steps, like a whimsical cook swapping salt for sugar. Scientists created a giant cookbook of real reaction chains and tested the AI with problems from a classic textbook — not a schoolbook, but the one that has trained organic chemists for decades, where a single misstep ruins the whole solution. The trained model began to follow recipes much more carefully, outperforming specialized software.

This skill will aid in drug discovery and materials design: instead of guessing, we'll get precise instructions for assembling the desired molecule step by step.