GPT-5.4 helped improve a hard drug-making reaction
GPT-5.4 was connected to Molecule.one’s Maria system to improve a difficult chemistry reaction used in drug research. The target was , a reaction that makes carbon-nitrogen bonds often needed in medicine-like molecules. This reaction is useful, but it can give low yields with some starting materials. The system created research ideas, designed experiments, read the data, and suggested tests.
Human chemists still guided the work, chose which ideas to test, corrected some lab plans, and independently checked the final result. The strongest idea was that an additive called TEMPO could improve the reaction between primary sulfonamides and boronic acids. Maria Lab ran 10,080 across two rounds. Under the best conditions, yields improved for 88% of tested boronic acids and 83% of tested sulfonamides.
Average yield rose from 16.6% to 25.2%, and above 30% yield rose from 15.6% to 37.5%. Human chemists repeated 14 selected reaction pairs at bench scale; 11 improved, and 8 improved by more than two times. The result is still early and needs independent and broader testing.
Key points
- GPT-5.4 worked with Maria to propose and plan chemistry experiments.
- Human chemists selected test ideas, fixed some plans, and results.
- TEMPO improved yields in a difficult reaction.
- Maria Lab ran 10,080 to test the idea across many combinations.
- The result is promising, but it still needs independent .