Google’s Antigravity agent helps biologists hunt for new antibiotics

Searching protein databases for new treatments is usually slow and split across many tools. In the walkthrough, the agent used ProtNLM, a Google model that predicts what a protein does from its sequence, to scan millions of uncharacterised proteins in UniProt. It then checked the best candidate’s novelty, its predicted 3D structure from AlphaFold and its neighbouring genes. The candidate targets bacteria behind respiratory infections and is now being tested in a wet lab.

Why it matters

The result is a lead, not a drug: it still has to pass lab tests. The bigger shift is in how the work gets done. Database triage that used to take days of scripts and file juggling becomes one guided session, so scientists can test more ideas and spend their time on the experiments that matter.