Industry story
Anthropic Launches Claude Science AI Workbench for Researchers
Anthropic is making a real bet that the bottleneck in scientific research is tool fragmentation, not ideas. Claude Science bundles databases, Jupyter notebooks, HPC cluster terminals, and 60-plus domain-specific skills — genomics, proteomics, cheminformatics — into one coordinated multi-agent environment, with a dedicated reviewer agent checking citations and calculations. Early results are striking: one neuroscientist cut a two-year review writing process to weeks. The question is whether that compression holds across disciplines, or whether it's concentrated in the life sciences where Anthropic has clearly focused its integrations first.
Full analysis
Anthropic has introduced Claude Science, a specialized AI workbench (an integrated software environment) aimed at scientific researchers. The app consolidates fragmented research tools — databases, Jupyter notebooks, R, HPC cluster terminals — into a single environment, featuring a coordinating agent with access to over 60 curated domain-specific skills covering genomics, proteomics, structural biology, cheminformatics, and more. Agents can spin up sub-agents, delegate to specialist agents, and a dedicated reviewer agent checks citations and calculations for accuracy.
Claude Science supports reproducible outputs by bundling figures with the code, environment, and message history that produced them. It manages compute scaling from a single GPU to hundreds via existing HPC clusters or Modal cloud compute, keeping sensitive data on-premises by only sending required context to Claude. The platform integrates NVIDIA's BioNeMo Agent Toolkit (a life-sciences model library) and is available in beta for Pro, Max, Team, and Enterprise Claude users on macOS and Linux. Early users include researchers at Manifold Bio and the Allen Institute, with one neuroscientist reporting a multi-agent pipeline that reduced scientific review writing from two years to weeks.
Comments