claudekit / updates / claude-science
[ NEW · ]

Introducing Claude Science

Anthropic launched Claude Science in beta on June 30, 2026, an AI workbench that integrates fragmented scientific research tools into a unified environment. Researchers can run complete workflows — from literature analysis through manuscript preparation — in one platform, with specialist agents that query and synthesize across more than 60 curated scientific databases including UniProt, PDB, Ensembl, and ChEMBL. It's available to Claude Pro, Max, Team, and Enterprise users on macOS and Linux.

Claude ScienceScience
Official announcement →

This article is a summary based on official documentation.

Overview

Anthropic launched Claude Science in beta on June 30, 2026. The tools researchers commonly use are scattered across separate environments, forcing constant context-switching; Claude Science integrates them into a single AI workbench that runs the full workflow from literature analysis through manuscript preparation. It serves scientists across life sciences disciplines including genomics, proteomics, structural biology, and cheminformatics.

Key features

  • Literature and data analysis

    Researchers had to move between databases and packages by hand for each study. Claude Science’s specialist agents query and synthesize across more than 60 curated scientific databases — including UniProt, PDB, Ensembl, and ChEMBL — and integrate the tools and packages researchers most commonly use.

  • Rich artifacts, fully reproducible

    Re-validating a result meant tracking down its code and environment separately. Claude Science natively renders 3D protein structures, genome browser tracks, and chemical structures, and every output includes complete code, environment details, and message history for validation.

  • Compute management

    Scaling an analysis meant provisioning compute and submitting and tracking jobs yourself. Claude Science manages your compute and scales on demand across personal laptops, HPC clusters, or GPU services via Modal, handling job submission and monitoring automatically.

  • Reviewer agent

    A built-in reviewer agent inspects the outputs, flagging incorrect citations, untraceable numbers, and figures that don’t match their underlying code.

Notes

  • Availability — available in beta for Claude Pro, Max, Team, and Enterprise users on macOS and Linux; Team and Enterprise admins must enable access
  • Use cases — single-cell RNA sequencing analysis, CRISPR screen design, protein structure prediction, long-form literature review generation, and germline variant analysis
  • Official announcement — anthropic.com/news/claude-science-ai-workbench
§ 4

Frequently Asked Questions

frequently asked
§ 4.1
What is Claude Science?
An AI workbench that integrates fragmented scientific research tools into a unified environment, letting researchers run complete workflows from literature analysis through manuscript preparation in one platform.
§ 4.2
Who is it for?
Scientists and researchers across life sciences disciplines including genomics, proteomics, structural biology, and cheminformatics, serving academic institutions, nonprofits, and industry research labs.
§ 4.3
Where is it available?
It's available in beta for Claude Pro, Max, Team, and Enterprise users on macOS and Linux. Team and Enterprise admins must enable access.
§ 4.4
How does it ensure trustworthy results?
Every output includes complete code, environment details, and message history, so it's fully reproducible. A built-in reviewer agent flags incorrect citations, untraceable numbers, and figures that don't match their underlying code.
§ 4.5
What can I use it for?
Use cases include single-cell RNA sequencing analysis, CRISPR screen design, protein structure prediction, long-form literature review generation, and germline variant analysis.