Anthropic has moved deeper into the laboratory. According to MarkTechPost, the company released Claude Science in beta on June 30, 2026, an application it bills as a multi-agent workbench for scientific research. TechRadar describes the tool simply as an "AI workbench" for scientists using Claude.
The design leans on teamwork between AI agents rather than a single chatbot. Per MarkTechPost, a coordinating agent hands tasks to domain specialists, while a separate reviewer agent flags and corrects citations and numbers. Crucially, the app is aimed at reproducibility: every figure it produces ships with its exact code, its computing environment, and the full message history behind it, so other researchers can retrace how a result was reached. MarkTechPost notes the tool runs on Anthropic's existing Claude models and targets genomics, proteomics, and cheminformatics pipelines.
The ambitions stretch beyond software. A report carried by finance.biggo.com says Anthropic launched the Claude Science Workbench and "vows to develop its own drugs." Crypto Briefing frames the initiative around drug discovery for neglected diseases—conditions that often draw little commercial research funding.
Taken together, the sources describe a company trying to position its AI not just as a writing or coding assistant but as an active participant in bench science, with built-in guardrails against the fabricated citations and shaky numbers that have dogged general-purpose chatbots.
Why it matters: if AI systems can make research pipelines faster while keeping every step auditable and reproducible, they could speed up work on diseases that have historically been overlooked—provided the results hold up to the scrutiny of real scientists.