OpenAI says its GPT-5.4 system, working with Molecule.one’s Maria lab platform, helped improve a tough medicinal chemistry reaction known as the Chan–Lam coupling. The work focuses on finding better conditions and additives to make the reaction produce more useful results.

Chan–Lam coupling is a common method chemists use to connect different chemical building blocks. In practice, it can be finicky, so even small gains in yield can matter when teams are trying to move from early experiments toward compounds they can test further.

According to OpenAI, Maria handled the wet-lab side, while GPT-5.4 supported the research process. The combined system generated research proposals, designed and ran experiments, analyzed the results, and then suggested follow-up experiments. OpenAI also says human chemists remained in the loop, which is a key point for how such systems are typically used in real labs.

OpenAI says Maria ran 10,080 reactions across two cycles. After that, the optimized conditions improved yields for a large share of the starting materials tested—88% of the boronic acids and 83% of the sulfonamides. The company presents this as evidence that the approach can systematically explore reaction options rather than relying only on one-off guesses.

Why this matters

For teams developing medicines, reaction reliability can be a bottleneck. If an AI-guided workflow can find higher-yield conditions for many related inputs, it can reduce trial-and-error time and help chemists spend more effort on the compounds that actually move forward.

Source: OpenAI