Google DeepMind · 2026-11-03 · notable
Antigravity Cheminformatics Skills — AI agent for drug-design analysis
Google DeepMind released Cheminformatics Skills for Google Antigravity. An agent gathers chemical data from papers and patents, runs matched molecular pair analysis, builds potency models and suggests synthesis routes.
Google Antigravity gets a Science plugin that lets an AI agent do early drug-design analysis.
What is it?
Cheminformatics Skills are a set of agent skills for Google Antigravity that let an agent collect compound data from literature and patents, curate it, and analyze it. The skills cover matched molecular pair analysis (MMPA), potency prediction models and synthesis route suggestions for new macrocycles.
How does it work?
The model plans the work and reasons about results, while deterministic chemistry tools built on RDKit do the actual calculations. In the demo, the agent studied crizotinib and its more rigid successor lorlatinib, then showed a dashboard of 568 curated compounds.
Why does it matter?
DeepMind says the full path from proposal to design took about an hour, compared with multiple days before. The authors also warn that sparse data and assay noise limit model precision, and that every design is a hypothesis that needs lab testing.
Who is it for?
medicinal chemists and drug-discovery teams
Try it
Download the Antigravity app and install the Science plugin