OpenAI and PerturbAI partner to accelerate drug discovery using AI agents

Biotechnology company PerturbAI has announced a collaboration with OpenAI and Scripps Research to develop an AI reasoning framework designed to accelerate drug discovery. The system utilizes approximately 2,000 AI agents working in parallel to analyze complex biological experimental data, retrieve scientific literature, and propose hypotheses for subsequent laboratory testing.

In a recent application involving PerturbAI’s whole-brain genome map, the multi-agent system completed data analysis and research integration in 32 minutes. The company estimates that performing such analysis and literature review through traditional human serial methods could take anywhere from several months to a year.

The AI agents function by analyzing cellular responses and linking experimental results with existing biological knowledge to create a “reasoning map.” This map connects genes, cellular reactions, and established research to provide scientists with leads for further investigation. In blind evaluations, biological experts preferred the explanations provided by the multi-agent system over those from single AI agents in 75% of cases.

Following this progress, the partnership intends to utilize causal biological data generated in living systems to focus on the discovery of new drug targets.