AI agents keep getting better at biology, but knowing when their scientific work deserves trust is a different problem. LatchBio co-founder and CTO Kenny Workman walks through how his company moved from biotech data infrastructure to benchmarking biological agents, starting with spatial biology and expanding into single-cell work, epigenetics, whole-genome sequencing and drug discovery. Speaking at a Greylock symposium, he explains why longer-horizon tasks matter and what would have to be true for AI systems to run substantial research autonomously.
Can anyone actually trust an AI scientist in the lab
LatchBio CTO Kenny Workman explains how his team turns real biology workflows into benchmarks that expose where AI agents succeed and fail.
