NVIDIA has introduced the BioNeMo Agent Toolkit, a software platform aimed at helping developers build AI agents that can carry out scientific workflows across biology, chemistry, genomics and drug discovery.
Announced at BIO 2026, the toolkit brings together NVIDIA’s life sciences models, libraries and accelerated computing technologies into a single platform, letting AI agents tap into scientific tools, analyze research data, run computational experiments and recommend next steps during the discovery process.
The company says the toolkit pushes AI platforms beyond simply answering scientific questions, enabling them to handle domain-specific tasks like protein structure prediction, molecular docking, genomic analysis, biomarker discovery and generative chemistry.
NVIDIA says the platform integrates technologies including BioNeMo, NIM microservices, Parabricks, NeMo, Nemotron, NemoClaw and OpenShell, creating an environment where AI agents can securely call scientific models, run workflows and interpret results.
More than 50 technology, pharmaceutical and research organizations are already adopting or integrating the toolkit, including Dassault Systèmes, Databricks, Eli Lilly, Schrödinger, Snowflake, and the University of Washington’s Institute for Protein Design. Anthropic and OpenAI are also folding BioNeMo capabilities into their own scientific workflows.
According to NVIDIA, organizations are using the platform for applications such as virtual drug screening, genomic analysis, protein design, biomedical research and medical imaging analysis. The company expects agent-based workflows to help researchers cut down the time needed for computational experiments while making scientific processes more consistent.
“Frontier models are the brains. BioNeMo is the scientific toolbox. Together, they give AI agents the skills of a PhD research assistant and the speed of a supercomputer,” said Jensen Huang, founder and CEO of NVIDIA. “For the first time, researchers can build AI agents that understand scientific knowledge, use scientific tools and execute scientific workflows.”
The University of Washington’s Institute for Protein Design is one of the research organizations working alongside NVIDIA to build out agent-ready scientific workflows. NVIDIA says the collaboration has helped roughly double runtimes for RosettaFold3 protein design models compared to the previous-generation implementation.
NVIDIA notes that BioNeMo is built to work with existing scientific software platforms, lab automation systems, cloud infrastructure providers and enterprise data environments — meaning organizations can slot AI agents into their current research workflows rather than having to replace the tools they already use.
The Bottom Line
Life sciences organizations are increasingly looking at AI agents that can handle multi-step scientific tasks, not just generate text or answer questions. NVIDIA’s BioNeMo Agent Toolkit reflects a wider industry move toward agentic AI platforms that combine reasoning with specialized scientific software — a shift that could shorten timelines for drug discovery, genomics research and other computational life sciences work.