Anthropic has unveiled Claude Science, an artificial intelligence (AI) workbench designed to provide scientists with a unified environment for computational research. Launched on Tuesday, the platform aims to eliminate the need for researchers to juggle multiple databases, pipelines, and tools, thereby streamlining complex scientific workflows. This strategic move by Anthropic signals a broader ambition to move beyond merely providing AI models and to establish itself as a key operating layer for specific industries.
Crucially, Anthropic clarifies that Claude Science does not introduce a new or more capable AI model. Instead, it runs on the same Claude models currently available, including Claude Opus 4.8, without any special access requirements. The workbench builds upon Anthropic's October 2025 launch of Claude for Life Sciences, which augmented the Claude chatbot's capabilities for life science tasks. Claude Science now offers a dedicated space to conduct this specialised work.
The functionality of Claude Science centres around a main AI assistant that acts as a project manager. This assistant can connect to more than 60 scientific databases and comes equipped with prebuilt toolkits tailored for specific fields such as genomics, protein structure, and chemistry. It can also create sub-assistants to delegate tasks or hand off work to custom 'expert' assistants built by users. A distinct fact-checker AI is integrated into the workflow to double-check citations and calculations before any findings are prepared for publication, addressing concerns about fabricated information in AI-assisted research.
Ensuring reproducibility is a core feature of Claude Science. The workbench can generate figures, such as 3D protein structures and chemistry diagrams, alongside the exact code and environment used to create them. Each figure includes a plain-language description of its creation and a full message history, allowing scientists to edit figures in plain language and prompt the agent to modify its underlying code. Furthermore, the platform can operate on a lab's own infrastructure, preventing the need to send sensitive data to Anthropic's servers and potentially saving valuable research time.
Early adopters have already reported tangible benefits. Jérôme Lecoq, a neuroscientist at the Allen Institute, utilised the tool to construct a multi-agent computational review pipeline. Similarly, Stephen Francis's team at the UCSF Brain Tumor Center leveraged Claude Science to drastically reduce the time required for comprehensive germline analysis of glioma, with their results independently validated. This launch positions Anthropic in a competitive landscape, with rivals like OpenAI and Google DeepMind also making strides in AI for scientific research, albeit with differing approaches.