Anthropic has unveiled Claude Science, an AI workbench designed to streamline scientific research by providing one unified environment for computational tasks. This platform connects with over 60 databases and comes equipped with prebuilt toolkits tailored to specific fields such as genomics, protein structure, and chemistry.

The system employs a hierarchical assistant architecture, where the main AI acts like a project manager, delegating tasks to sub-assistants or even custom expert assistants. A fact-checking AI ensures accuracy before any findings are published.

Anthropic’s approach stands in contrast to OpenAI's GPT-Rosalind and Google DeepMind’s Gemini for Science, which both use specialized models but with different distribution strategies. This diversity signals a shift in how AI vendors will compete in niche scientific markets.

The beta launch comes with clear criteria: 50 projects awarded through an application process, focusing on biomedical research and other interdisciplinary fields. These initiatives aim to explore the frontiers of science using advanced AI tools.