Biolevate Integrates NVIDIA BioNeMo to Scale Agentic Drug Discovery

The integration aims to move beyond simple information retrieval, allowing Biolevate’s platform to perform mechanistic inference. By utilizing NVIDIA’s GPU-optimized pipelines and model-training architectures, the startup intends to refine its RNA-focused models, specifically targeting improvements in predicting non-coding RNA folding and hybridization. These high-fidelity models are expected to enhance patient stratification and streamline the drug discovery process.

Biolevate’s CEO Joël Belafa noted that the future of life sciences lies in systems that reason across evidence and experimentation. By aligning its OCEAN stack with NVIDIA Nemotron and BioNeMo open-source models, the company plans to reduce computational friction and increase the interoperability of its research environment with the broader technological ecosystem. This collaboration builds on Biolevate's existing use of accelerated computing, which the company already employs to achieve high accuracy in clinical and regulatory documentation tasks. The platform currently supports major industry players, including Sanofi and Elsevier, in bridging the gap between computational predictions and laboratory validation.