Claude generates complex molecule synthesis routes using code alone

Anthropic's Claude has generated complete synthesis routes for complex molecules using only code, without specialized chemistry software or databases. The model produced step-by-step reaction sequences for several drug-like targets, demonstrating that general-purpose AI can perform retrosynthesis—working backward from a desired molecule to affordable starting materials. This matters because retrosynthesis is a bottleneck in drug discovery, and faster route-finding could compress years of medicinal chemistry work into hours. Accuracy remains the key open question: the article notes that proposed routes still require wet-lab validation, and errors in reaction feasibility could limit near-term adoption. Watch for benchmarking against expert chemists and integration into pharma discovery pipelines as the next signals of real-world utility.