Google Antigravity adds Cheminformatics Skills for chemistry research

What is changing?

Google DeepMind released Cheminformatics Skills for Antigravity on November 3. In the company’s demonstration, an agent collected thousands of compounds from scientific literature and catalogs, filtered irrelevant records, ran matched molecular pair analysis and built predictive models. It then proposed new rigidified macrocycles and possible synthesis routes.

The language model handles workflow orchestration, database navigation and scientific reasoning, while deterministic tools execute molecular calculations. The skills are grounded in RDKit and use connectors to ChEMBL and PubChem. DeepMind says the demonstrated workflow completed one iteration in about an hour, compared with a process that could previously take multiple days.

This is not a drug discovery result or laboratory validation. The company explicitly notes that sparse data, inter-lab assay noise and narrow dynamic ranges limit predictive precision. Generated designs are testable hypotheses that must be synthesized and experimentally validated. The skills are available today through Antigravity’s Science plugin and enterprise accounts.