AI Chemist Lifts Chan-Lam Yields to 25.2% Across 10,080 Reactions
Updated
Updated · OpenAI · Jun 17
AI Chemist Lifts Chan-Lam Yields to 25.2% Across 10,080 Reactions
1 articles · Updated · OpenAI · Jun 17
Summary
OpenAI and Molecule.one said a near-autonomous system improved a hard Chan-Lam coupling reaction, raising mean yield to 25.2% from 16.6% for primary sulfonamides with boronic acids.
GPT-5.4 identified mild oxidants including TEMPO as the key idea, then worked with Maria AI and a high-throughput lab through two experimental cycles that ran 10,080 reactions in three months.
Under optimized conditions, yields improved for 88% of boronic acids and 83% of sulfonamides tested, while the share of reactions above 30% yield rose to 37.5% from 15.6%.
Bench-scale repeats backed the microliter results, with human chemists confirming higher yields in 11 of 14 substrate pairs and more than doubling yield in most cases.
The team called the workflow near-autonomous because humans still selected proposals, corrected plans and validated results; independent replication and broader substrate testing are next.
Two years on, has AI chemistry discovered a truly novel reaction, or just optimized what humans already know?
Has this AI-driven breakthrough led to any new drug candidates entering clinical trials in the past two years?
As AI generates endless ideas, is the real bottleneck now human insight or the physical limits of robotic labs?
AI-Driven Breakthrough in Chan-Lam C–N Coupling: High-Throughput Experimentation Enables Predictable, Broad-Scope Mono-Arylation of 3,904 Sulfonamides (2025–2026)
Overview
This report highlights how advanced computational methods, especially machine learning and high-throughput experimentation, are transforming C–N coupling reactions such as the Chan-Lam coupling, which are essential for pharmaceutical discovery and synthesis. By applying these technologies, researchers have broadened the scope and improved the performance, efficiency, and predictability of these complex reactions. The Machine Learning for Pharmaceutical Discovery and Synthesis (MLPDS) Consortium plays a key role by developing software to automate small molecule discovery. Notably, AI-driven approaches now enable chemists to tackle challenging substrates like primary sulfonamides, overcoming limitations of traditional methods and accelerating progress in drug development.