Anthropic Launches Claude Science Workflow for Researchers
Original: Anthropic’s Claude Science bets on workflow, not a new model, to win over scientists
Why This Matters
Demonstrates shift from model-centric to workflow-based AI competition in scientific research applications.
Anthropic introduced Claude Science on June 30, 2026, an AI workbench designed for scientists to conduct computational research in a unified environment. The tool integrates with over 60 scientific databases and uses existing Claude models without requiring new capabilities or special access.
Anthropic unveiled Claude Science on Tuesday at an AI for Science briefing, positioning it as a dedicated workbench rather than a new AI model. The platform runs the same Claude models available to all users, including Claude Opus 4.8, with no special gating or enhanced capabilities for biology. Claude Science builds on Anthropic's October 2025 launch of Claude for Life Sciences, which enhanced the Claude chatbot for life sciences tasks. The workbench features a main AI assistant functioning as a project manager that connects to more than 60 scientific databases and includes prebuilt toolkits for genomics, protein structure, and chemistry. The assistant can create sub-assistants to delegate work or hand tasks to custom expert assistants built by users. A separate fact-checker AI verifies citations and calculations before publication to prevent fabricated citations from entering papers. Claude Science generates reproducible research figures including 3D protein structures and chemistry drawings alongside the underlying code that created them, with plain-language descriptions and full message histories. Scientists can edit figures using natural language, prompting the agent to modify underlying code automatically. The workbench runs on laboratory infrastructure rather than sending data to Anthropic's servers, addressing data privacy concerns. Anthropic positions Claude Science as part of its broader strategy to provide vertical, workflow-level products for specific industries, similar to Claude Code's role in software development, moving beyond pure model capability competition.