Anthropic Launches Claude Science, an AI Workbench Built for Researchers
Anthropic has launched Claude Science, an AI workbench for scientists that integrates the tools and packages researchers most commonly use, produces auditable artifacts, and provides flexible access to computing resources. The new app is now available, and it aims to fix one of the most frustrating parts of research work: jumping between dozens of disconnected tools just to get anything done.
According to Anthropic, researchers typically work across dozens of databases with different schemas, deal with file formats that need custom data pipelines and viewers, and switch between tools like PubMed, Jupyter, R, and cluster terminals. Claude Science brings all of that into one place. It helps analyze literature, run multistep research, produce detailed artifacts, and refine figures and manuscripts until they are publication ready, with every output carrying an auditable history so results can be validated and reproduced.
Inside Claude Science, users work with a generalist coordinating agent with access to over 60 curated skills and connectors preconfigured for genomics, single cell research, proteomics, structural biology, cheminformatics, and more. These agents can also launch specialist sub agents, and a reviewer agent checks citations and calculations, flagging and correcting errors along the way.
One standout feature is how the platform handles compute heavy tasks. For large analyses like protein folding or genomics pipelines, Claude Science drafts a plan, asks before reaching new resources, and lets users review or revoke decisions before submitting jobs to their lab’s own HPC cluster or to Modal for on demand compute, scaling from a single GPU to hundreds as needed. Sensitive datasets never have to leave existing systems since the tool runs locally or on a lab’s own infrastructure.
Claude Science also comes domain ready from day one. It taps into NVIDIA’s BioNeMo Agent Toolkit to connect natively with life sciences models and libraries including Evo 2, Boltz-2, and OpenFold3.
Early users are already seeing results. Jérôme Lecoq, a neuroscientist at the Allen Institute, used Claude Science to build a multi agent review pipeline that previously took his team up to two years to complete manually; he now has about 10 completed reviews, several over 100 pages long. Meanwhile, Stephen Francis, an epidemiologist at the UCSF Brain Tumor Center, said the tool sped up his glioma research workups to roughly one tenth of the time it used to take, with results independently validated by his team.
For now, Claude Science is available in beta on macOS and Linux for Pro, Max, Team, and Enterprise plans. Anthropic is also opening applications for its AI for Science program, offering up to 50 projects a combined $30,000 in credits, with Modal providing up to $2,000 in compute for select projects; applications close July 15, 2026.
