Anthropic has introduced Claude Science — a new AI-based research workbench designed to simplify the management of complex research workflows by unifying data access, computation, and analysis in a single environment.
The beta version of the product is available to users with Pro, Max, Team, and Enterprise subscriptions and aims to reduce the fragmentation researchers face when switching between separate tools and databases.
The platform integrates with more than 60 scientific databases and includes pre-built toolkits adapted for disciplines such as genomics, proteomics, structural biology, and chemoinformatics. At the core of Claude Science is a multi-agent system in which a main agent coordinates tasks among specialized sub-agents or user agents, distributing workloads according to research requirements.
A key feature of the system is a specially designed verification agent that independently audits findings before publication. This layer is intended for citation checking, calculation validation, and reducing the risk of error propagation in research processes. Anthropic explained that Claude Science does not rely on a new model, but runs on existing Claude models, including Opus 4.8.
The workbench can be deployed in multiple environments, including local macOS and Linux systems, remote servers accessible via SSH, and high-performance computing (HPC) clusters. For large-scale computational tasks such as protein folding modeling or genomic pipelines, the system can automatically submit jobs to institutional HPC infrastructure or cloud services such as Modal, scaling from single-GPU workloads to distributed computing environments.
Anthropic emphasizes reproducibility as a key design principle. Every generated result is packaged with the underlying code, computational environment, methodological explanation, and full interaction history, enabling other researchers to reproduce the results. Users can also modify outputs using natural instructions, and the system automatically adjusts the code accordingly.
Early users, including researchers from the Allen Institute and the UCSF Brain Tumor Center, reported significant efficiency improvements. In some cases, literature analysis processes that previously took months or even years were reduced to much shorter cycles, and complex genomic analyses were completed in a fraction of the usual time with automated verification support.
The launch builds on Anthropic's expanding focus on biological sciences, following its partnership with Bristol-Myers Squibb and the earlier release of Claude for Life Sciences. The company has also launched an internal drug development initiative targeting diseases that are often overlooked by traditional pharmaceutical economics, indicating a deeper entry into applied scientific research.
To encourage adoption, Anthropic is offering grants of up to $30,000 in Claude Science credits for selected research teams, with applications accepted until July 15, 2026.