Anthropic expands Claude for biomedical research

Anthropic is turning Claude into a tool that covers more stages of biomedical research: from reviewing studies and analyzing genetic data to preparing protocols, clinical documents, and regulatory materials.

The company introduced Claude for Life Sciences, a set of improvements, connections to scientific platforms, and specialized resources for researchers, trial coordinators, and regulatory affairs teams. The goal is for Claude to support more than isolated tasks and assist throughout the development of a drug.

A model with better results on scientific tasks

Anthropic says its Claude Sonnet 4.5 model outperforms earlier versions on life sciences evaluations. In Protocol QA, a test of laboratory protocol comprehension and application, it scored 0.83, compared with a 0.79 human benchmark and 0.74 for Claude Sonnet 4.

The company also says it has improved performance on BixBench, an evaluation of bioinformatics tasks, the field that uses software and data analysis to study biological information.

These figures do not mean Claude can replace a researcher or validate an experiment on its own. They indicate that it can interpret technical instructions more effectively and help with specific parts of the work, always under human review.

Claude connects to laboratory tools

The main development is not just the model itself. Anthropic is adding connectors so Claude can access data and work alongside platforms scientists already use:

• Benchling: answers questions by linking to experiments, lab notebooks, and original records.
• BioRender: provides access to a library of reviewed scientific figures, icons, and templates.
• PubMed: makes it easier to search millions of biomedical articles and clinical studies.
• Wiley's Scholar Gateway: provides access to peer-reviewed scientific content.
• Synapse.org: enables data sharing and analysis across public or private projects.
• 10x Genomics: offers single-cell and spatial data analysis through natural-language instructions.

Claude can also connect to enterprise services such as Google Workspace, Microsoft SharePoint, OneDrive, Outlook, and Teams. For large-scale analysis, Anthropic mentions integrations with Databricks and Snowflake.

In practice, this makes it possible to ask Claude to analyze a genetic dataset, find the studies supporting a hypothesis, and present the results in a document or presentation. Its usefulness depends on permissions, data quality, and team oversight.

Reusable protocols for repetitive tasks

Anthropic is also developing Agent Skills, folders containing instructions, scripts, and resources that Claude can follow to perform a task more consistently.

The first highlighted scientific skill is single-cell-rna-qc, designed to review and filter single-cell RNA sequencing data according to recommended scverse practices. Scientists can also create custom skills for their internal procedures.

This can be useful when a lab repeats the same process frequently. Instead of explaining each time how to clean the data or apply a protocol, the team can prepare those instructions in advance and have Claude follow them more consistently.

Tasks Claude can support

Anthropic outlines several uses for Claude in the life sciences:

• Review scientific literature, summarize articles, and propose testable hypotheses.
• Draft protocols, standard operating procedures, and consent documents using data connected to Benchling.
• Analyze genomic information and produce results as code, documents, presentations, or analysis notebooks.
• Prepare and review documentation for clinical and regulatory processes.

The company is also preparing a library of instructions to help users get better results without having to learn how to write complex prompts.

Companies and labs are already testing it

Anthropic says organizations including Sanofi, Broad Institute, 10x Genomics, Schrödinger, Komodo Health, FutureHouse, and Axiom already use Claude for various scientific or business tasks. The examples mentioned include literature analysis, software development, drug toxicity prediction, and clinical data exploration.

These are customer and partner testimonials, not independent results showing that Claude accelerates every project equally. In regulated sectors, any output used to make decisions must go through quality controls, traceability checks, and expert review.

Claude for Life Sciences is already available through Claude.com and AWS Marketplace. Anthropic says it will soon arrive on Google Cloud Marketplace and that it offers API credits to researchers through its AI for Science program.

The important change is the shift in approach: Anthropic is not presenting Claude only as an assistant for writing code or summarizing articles, but as a layer that connects AI models with research data, protocols, and tools. The next point to watch is whether that integration produces measurable advances in real laboratories without compromising accuracy, safety, or regulatory accountability.