The Antibiotic Scientist
an AI running on cloud, designing new medicine for a drug-resistant infection
connecting…
the target
Staphylococcus aureus
DNA gyrase B
This is the bacterial enzyme the AI is attacking. It's what the MRSA superbug uses to replicate its DNA — block it, and the bacteria can't grow. Every molecule below is a proposal for something that fits into this protein's ATP pocket.
loading structure…
3D · drag to rotate · scroll to zoom
loading…
—
Reading live data from the agent…
—
experiments run
02
how it works
We bill for working,
not waiting.
Watch this agent live through its current session. Each block is a slice of time.
WORKING
start
now
in a run
between runs (checkpointed)
—
—
session length
wall clock since first run
—
time in Claude runs
from agent log
—
checkpoint cycles
Orb froze / woke the container (30-day total)
experiments
Each molecule is tested by twelve independent filters.
If all twelve pass, the candidate is published. If any one fails — you'll see why below.
loading experiments…
What you're actually watching.
Each card is a molecular design experiment. The AI proposes a new compound, then runs it through a twelve-step filter chain: chemistry sanity, binding simulation against the bacterial target, toxicity prediction, novelty check against all known drugs, and an adversarial review. Only candidates that survive all twelve are flagged as leads.
These are computational hypotheses, not therapies. They have not been synthesised or tested in any biological assay.