NVIDIA and Anthropic Bring AI Agents to Scientific Research With Claude Science
NVIDIA and Anthropic have announced a new collaboration that aims to transform scientific research by combining advanced AI agents with GPU-accelerated computing. Anthropic’s new Claude Science platform now integrates with the NVIDIA BioNeMo Agent Toolkit, allowing researchers to perform complex life sciences workflows using simple natural language commands.
The integration marks another step toward agentic AI for science, where AI systems can reason, plan, and execute specialized scientific tasks with minimal manual setup.
Natural Language Meets Scientific Computing
Claude Science is designed as an AI-powered workbench that allows researchers to describe scientific tasks in plain language instead of manually configuring software, AI models, or computing infrastructure.
Through its integration with NVIDIA BioNeMo Agent Toolkit, Claude Science can automatically select the appropriate NVIDIA-accelerated tools, prepare workflow inputs, execute computational tasks, and return results within the same research environment.
This approach allows scientists to stay focused on research rather than managing complex computational pipelines.
Faster Drug Discovery and Genomics Research
The platform supports a wide range of life sciences applications, including:
- Drug discovery
- Genomics
- Protein structure prediction
- Molecular design
- Protein engineering
- Single-cell analysis
- Clinical research
- Cheminformatics
Scientists can ask Claude Science to perform tasks such as analyzing genomic sequences, predicting protein structures, or designing therapeutic molecules. The AI coordinates specialized domain agents while NVIDIA’s accelerated computing stack handles the computational workload.
One highlighted example involves cancer research. A scientist can provide a known cancer-causing mutation and ask Claude Science to generate and optimize potential inhibitors. NVIDIA BioNeMo accelerates prediction, optimization, and validation of these candidate molecules, significantly reducing the time required for computational screening.
NVIDIA Technologies Behind the Platform
NVIDIA says its BioNeMo Agent Toolkit connects AI agents with several accelerated scientific tools, including:
- NVIDIA Parabricks, which reduces genomic analysis from hours to minutes.
- RAPIDS-singlecell, developed with scverse, capable of processing a 1.3-million-cell workflow in about 25 seconds instead of 52 minutes.
- nvMolKit, which accelerates cheminformatics operations such as similarity searches and conformer generation by up to 3,000 times.
- BioNeMo open models for biomolecular AI workloads.
- BioNeMo NIM microservices, providing enterprise-ready inference endpoints for production deployments.
These tools enable AI agents to rapidly perform specialized scientific computations while maintaining high performance.
Broad Adoption Across the Pharmaceutical Industry
NVIDIA also revealed that 18 of the world’s top 20 pharmaceutical companies already use BioNeMo technologies across various AI-driven research initiatives.
The platform has become an important component of modern computational biology, supporting research in genomics, medical imaging, molecular simulation, and drug development.
Open and Flexible Platform
The BioNeMo Agent Toolkit is designed to be open and compatible with multiple AI agent frameworks rather than being limited to a single ecosystem. Developers can access the toolkit and its scientific skills through NVIDIA’s developer resources and GitHub.
Meanwhile, Anthropic has launched Claude Science into public beta, inviting researchers to test the platform and provide feedback on additional scientific integrations and domain-specific capabilities.
As AI agents become increasingly capable of handling sophisticated scientific workflows, collaborations like this could significantly shorten research cycles and accelerate discoveries in medicine, biotechnology, and life sciences.