Anthropic unveils Claude Science platform and trims Claude Code prompts

Anthropic announced Claude Science, an AI workbench for scientific research, at a San Francisco event on June 30. The beta‑stage platform integrates over 60 scientific skills and connectors—covering genomics, single‑cell analysis, proteomics, structural biology and chemistry—to let researchers query data, run multi‑step analyses, generate visualizations and retain reproducible records. Anthropic also revealed an internal pre‑clinical drug‑discovery effort targeting diseases that receive limited attention from traditional biopharma, with its CEO estimating the tool could shorten candidate‑to‑approval timelines from about 12 years to 7‑8 years.

In a separate development, Anthropic reduced the system‑prompt length for its Claude Code coding assistant by roughly 80 percent, arguing that fewer built‑in prohibitions unlock model creativity. The change, highlighted by Anthropic engineer Talik Shipar at a July 1 engineering event, removes many preset behavior examples and emphasizes giving the model appropriate tools to solve tasks, such as writing scripts to filter data.

Both moves signal Anthropic’s push to broaden AI utility beyond chat, extending into scientific workflows and more flexible programming assistance.