The Point: Anthropic is systematically working to optimize Claude for standard chemical tasks such as NMR spectral analysis to relieve chemists of time-consuming work with various molecular representations.
Anthropic is advancing Claude specifically for chemical tasks and publishing initial results on NMR spectral processing. The company is using collaboration with specialist chemists in synthesis, computational simulation, and analytics.
Chemists work daily with multiple representation forms of the same molecules: hand-drawn structures, instrument outputs, database queries, and technical notations from patents and publications. Each of these forms encodes the same chemistry but requires different skills. A structural sketch, for example, allows similarities to be recognized and predictions to be made; but it does not help distinguish between very similar molecules.
Correct molecular identification is central: tiny changes in structure can turn a medicine into a poison, as the thalidomide disaster demonstrates. The CAS Registry, the world’s largest chemical database, lists over 290 million substances with approximately 15,000 added daily. Manual translation between different representation forms is time-consuming and not scalable.
Previous machine learning approaches focused on retrosynthesis and reaction prediction but often failed due to fragmented, inconsistent data and paywalls. The new multimodal frontier models like Claude are changing the situation: they can read chemical structures directly from sketches or journal article images without relying on pre-built databases. They can also process experimental details in the form they are actually published and present their reasoning transparently.
Anthropic has published initial research testing Claude on one of the most common analytical tasks performed by chemists: the analysis of NMR spectra (nuclear magnetic resonance spectroscopy). The work compares Claude’s performance with established chemistry software and shows that the model is beginning to meaningfully support chemists in everyday translation, recall, and integration work.
Source: www.anthropic.com · Published June 5, 2026
Lumi AI News — AI-assisted curation pursuant to Art. 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.6.5.