Anthropic has introduced Claude Science, an AI workbench designed to streamline and accelerate scientific research by integrating various laboratory tools into a single platform. This new app enables AI agents to take over significant portions of experimental work, reducing the need for human intervention in time-consuming tasks. The launch marks Anthropic's most significant foray into the laboratory setting, aligning with its broader strategy to position Claude as a tool capable of real research, not just conversation.
Claude Science has already demonstrated impressive capabilities in protein design and analytical chemistry. In protein design, Claude (using Mythos Preview and Opus 4.8) successfully designed protein binders against 14 out of 15 targets, achieving hit rates of 26.7% and 22.6% respectively, compared to the typical 10-15% in current campaigns. For analytical chemistry, Claude Opus 5, working within Claude Science, processed raw NMR and LC-MS files in 23 and 19 minutes, matching the lab's own processing results and significantly reducing the time a chemist would typically spend on such analyses, which can range from 30 minutes to several hours per sample.
The platform centralizes scattered tools, databases, and file formats that researchers typically navigate, bringing them into a unified research environment. It offers features such as auditable artifacts, flexible computing resource access, and a generalist coordinating agent with over 60 pre-configured skills for various scientific domains like genomics, proteomics, and cheminformatics. Claude Science manages compute needs, drafting plans and scaling analyses on demand, from a single GPU to hundreds. It is currently available in beta for Claude Pro, Max, Team, and Enterprise users on macOS and Linux, with Anthropic supporting up to 50 AI for Science projects with up to $30,000 in credits, and Modal providing up to $2,000 in compute for select projects.
