Anthropic, NVIDIA Move AI Agents Much Deeper into Scientific Workflows
- By John K. Waters
- 07/01/26
Anthropic has presented Claude Science, a new AI workbench for researchers that integrates research tools, produces auditable artifacts, and connects to specialized life sciences designs and workflows from NVIDIA.
The beta release, revealed June 30, provides Claude Pro, Max, Group, and Enterprise users access to an application created to help researchers work throughout literature evaluation, information analysis, figure generation, manuscript improvement, and computational workflows. Anthropic said Claude Science is available on macOS and Linux, and can run locally, on remote machines over SSH, or through a high-performance computing login node.
The launch belongs to a wider push by AI companies to turn general-purpose assistants into domain-specific workbenches for professional users. In life sciences, that means moving beyond chat-based summarization towards representatives that can query databases, compose and run code, examine outputs, maintain research study history, and connect to scientific tools already utilized by labs.
Anthropic said Claude Science brings fragmented scientific tools into a single research environment. The app can work with tools such as PubMed, Jupyter, R, cluster terminals, and domain-specific scientific databases, while protecting an auditable history of how outputs were produced.
Users communicate with a generalist collaborating representative that has access to more than 60 curated abilities and ports set up for genomics, single-cell analysis, proteomics, structural biology, cheminformatics, and other research study locations. The system can also utilize expert agents developed by users, and consists of a reviewer agent that inspects citations and calculations, flagging and correcting mistakes, according to Anthropic.
A main claim in the launch is reproducibility. When Claude Science creates a figure, it consists of the code and environment utilized to develop it, a plain-language description of the procedure, and the message history leading up to the output. The company stated that history is intended to make results much easier to verify and replicate later on.
That focus is likely to matter for researchers and enterprise AI purchasers. Scientific AI tools might be evaluated less by whether they can produce possible responses and more by whether their work can be traced, challenged, repeated, and integrated into recognized review processes.
NVIDIA’s function comes through its BioNeMo Representative Toolkit, which the chipmaker revealed June 23. Anthropic said Claude Science utilizes BioNeMo Agent Toolkit skills to connect to life sciences designs and libraries in BioNeMo, including Evo 2, Boltz-2, and OpenFold3.
NVIDIA describes BioNeMo Representative Toolkit as a set of domain-specific tools and skills for agentic life sciences workflows. The toolkit includes NVIDIA life sciences libraries, tools, and open models, and is created to help representatives collect proof, reason throughout findings, run computational experiments, and advise next steps.