Anthropic Launches Claude Science, an AI Workbench for Researchers
Claude Science consolidates the multi-tool workflow scientists typically use into one verifiable environment, potentially reducing friction in research and publication.
Key Facts
- Anthropic released a beta version of Claude Science on Tuesday, an AI workbench for scientific research.
- The app integrates literature search, data analysis, chart and manuscript creation, and computing resource management into a single interface.
- Claude Science comes pre-configured with over 60 skills and connectors for fields including genomics, cell analysis, protein analysis, structural biology, and cheminformatics.
- The app includes a review agent that detects incorrect citations, numbers that cannot be traced to sources, and charts that do not match the code, then attempts corrections.
- Claude Science is available on macOS and Linux for Pro, Max, Team, and Enterprise plans, with discounted academic and non-profit pricing.
Reporting from 2 sources: ASCII.jp, GIGAZINE.
Anthropic released a beta version of Claude Science, an AI workbench that integrates literature research, analysis, chart and manuscript creation, and computing resource management into a single environment. It comes with over 60 pre-configured skills for fields like genomics and structural biology, supports HPC and cloud computing, and includes a review agent for detecting errors.
Anthropic released a beta version of Claude Science on Tuesday, an AI workbench designed for scientific research. The app integrates literature search, data analysis, chart and manuscript creation, and computing resource management into a single interface. It comes pre-configured with over 60 skills and connectors for fields including genomics, cell analysis, protein analysis, structural biology, and cheminformatics. The orchestrator agent can launch specialized agents and collaborate with user-created ones.
Claude Science saves the code used to generate charts, the execution environment, plain-language explanations, and conversation history together, allowing researchers to verify or reproduce results later. Users can issue natural-language instructions like removing grid lines or changing axis scales. For computation, the app connects to a lab's HPC cluster via SSH or on-demand resources through a Modal account, scaling from a single GPU to hundreds. Large-scale data stays on the lab's local machine or HPC node, with only the necessary context sent to Claude.
The app includes a review agent that detects incorrect citations, numbers that cannot be traced to sources, and charts that do not match the code, then attempts corrections. Specialized agents search across databases such as UniProt, PDB, Ensembl, Reactome, ClinVar, ChEMBL, and GEO, and utilize skills from NVIDIA's BioNeMo Agent Toolkit, including models like Evo 2, Boltz-2, and OpenFold3. Claude Science is available on macOS and Linux for Pro, Max, Team, and Enterprise plans, with discounted academic and non-profit pricing and a program supporting up to 50 AI for Science projects.
Synthesized by Yomimono from the 2 cited sources below, including Japanese-language reporting where cited, then editorially reviewed before publishing.