Anthropic’s Claude Science bets on workflow, not a new model, to win over scientists

Key Points:

• Anthropic launched Claude Science, an AI workbench designed to streamline computational research by integrating over 60 scientific databases and prebuilt toolkits into a single environment for scientists.
• Claude Science uses existing Claude models without special enhancements, featuring a main AI assistant that manages projects, delegates tasks to sub-assistants, and includes a fact-checker AI to verify citations and calculations before publication.
• The platform emphasizes reproducibility by generating figures alongside the code and environment that created them, allowing scientists to edit visuals through plain-language prompts, and can operate on the lab’s local infrastructure to protect data privacy.
• Early adopters like the Allen Institute and UCSF Brain Tumor Center report significant time savings and enhanced research workflows using Claude Science for complex analyses.
• Claude Science competes with OpenAI’s GPT-Rosalind, which is enterprise-gated and fine-tuned for biology, and Google DeepMind’s proprietary foundational science models, highlighting differing strategies in the AI scientific research market; Claude Science is currently in beta for various subscription tiers and offers grants for select biomedical research projects.