Anthropic launches Claude Science: AI workspace for researchers

Anthropic released Claude Science, an AI workbench built for scientists that pulls together dozens of databases, tools, and packages into one interface and offers built-in workflows for analysis, charting and manuscript drafting. The app is available in beta to Pro, Max, Team, and Enterprise users and includes an internal verification agent that checks citations and calculations, Anthropic says.

What is Claude Science?

Claude Science is an integrated AI workspace for research that provides more than 60 preconfigured skills spanning genomics, proteomics and cheminformatics, lets researchers analyze literature, run multi-step analyses, create charts and draft manuscripts, and includes a verification agent to validate citations and calculations autonomously. The app also allows users to save custom pipelines as reusable skills for repeated workflows.

Beyond those core features, the primary launch description emphasizes that the app bundles dozens of external databases, domain-specific tools and packages into a single interface so scientists do not have to stitch separate systems together for typical research tasks.

How does Claude Science handle data and compute?

The app runs locally on macOS or Linux and connects to remote machines through SSH or HPC clusters so sensitive data can remain inside a lab's infrastructure; "Only the context Claude actually needs gets sent to the model," Anthropic says, and jobs can scale from a single GPU to hundreds when extra power is required. Claude Science therefore supports hybrid workflows where the interface and orchestration live on local machines while heavy compute can be delegated to larger clusters.

Claude Science also integrates Nvidia's new BioNeMo agent toolkit, which ships with models such as Evo 2, Boltz-2 and OpenFold3, giving researchers ready access to tooling the company positions for computational biology and structural modeling workloads.

How does Claude Science support scientific workflows and validation?

Researchers can use the workspace to run literature reviews, chain multi-step analyses, visualize results and draft manuscripts, while saving successful sequences as reusable skills to standardize methods across projects. The workspace includes a verification agent that checks citations and calculations on its own, and Anthropic says users can plug in dozens of external databases and packages without transferring all raw data out of their infrastructure.

At an event for pharmaceutical executives, biotech founders and researchers, where the product was shown to industry guests, MIT Technology Review noted Claude Science was presented as a counterpart to Claude Code but focused on scientific research. The same outlet reported Anthropic plans to use Claude Science in its own internal drug research into rare and neglected diseases.

Why it matters

Centralizing domain tools and adding an autonomous verification agent addresses two persistent frictions in computational biology: costly tool integration and the risk of unchecked outputs. Anthropic is positioning Claude Science as a platform that keeps sensitive data on-premises while still enabling large-scale compute, a combination likely to interest labs and pharmaceutical groups that must control data flows and auditability.

MIT Technology Review described the launch as a sign Anthropic is doubling down on AI for science, and Anthropic is offering financial support to accelerate initial adoption, which increases the odds the product will see real-world workflows rather than only demos.

What to watch

Claude Science is in beta for Pro, Max, Team and Enterprise users, and Anthropic is backing up to 50 research projects with up to $30,000 in credits each; applications for that program are open through July 15, 2026. Watch which academic labs and pharma groups take the credits, and whether Anthropic’s own drug research using Claude Science produces preprints or public demonstrations of reproducible pipelines.