Google DeepMind has published a walkthrough of Cheminformatics Skills, a new set of capabilities for its Google Antigravity app, dated November 3, 2026. The skills are available now. Users install the Science plugin in the Antigravity app, and Enterprise users can choose "Use Business Account" and log in with Google Cloud credentials.
What the agent does
The demonstration centers on lung cancer drugs. The agent searches literature for crizotinib, lorlatinib and related compounds, extracts the molecules, links them to the reported data and discards irrelevant entries such as solvents and buffers. It retrieves thousands of compounds within minutes and writes a report that documents where each came from.
From there it opens an interactive viewer with scatter plots, runs matched molecular pair analysis, and benchmarks several machine learning models on four experimental readouts. It then proposes new rigidified macrocycles, lists synthesis routes, draws reaction mechanisms and predicts potency for each design. Google says the whole run takes about an hour, where a single iteration would normally take multiple days.
How it is built
The skills split the labor. The language model handles orchestration, database navigation and scientific reasoning. Deterministic chemistry tools do the molecular calculations. Those tools rest on the open source RDKit toolkit and connect to ChEMBL and PubChem. Google's reasoning is that a single changed atom or bond describes a different compound, so calculations cannot be left to the model.
Limits Google states
The post says sparse data, assay noise between labs and narrow dynamic ranges cap how precise the models can be. The proposed designs are hypotheses, and each one still has to be synthesized and tested in a lab.
Access
Google is offering the highest usage tier of Antigravity free for 6 months to 10,000 scientists, onboarded on a rolling basis and subject to eligibility, capacity and selection. Applications are open.
Top comments (0)