Building an AI-first enterprise: real-world examples

Moving from isolated pilots to AI-native operations means rethinking entire workflows around AI rather than bolting on point solutions. The most successful programs share a pattern: a breakthrough in one function becomes the catalyst for adoption across the enterprise. From day one, Anthropic has been committed to working with life sciences organizations to help them realize the potential of AI across their organization. In fact, earlier this year, we announced our partnership with Bristol Myers Squibb to deploy Claude across the company's research, clinical development, manufacturing, commercial, and corporate functions.

The examples below show how organizations use Claude across the value chain, and how early wins create momentum for broader change.

Research and discovery: from literature to lab

Life sciences organizations see their biggest gains when they connect related workflows across discovery and development.

Bioinformatics

Biomni (opens in new tab) set out to remove the bottleneck that locks most scientists out of genomic insight: the deep programming expertise that bioinformatics pipelines usually require.

The challenge: specialist dependencies slowing science

A bench scientist with a hypothesis often can't run a genomic analysis without a bioinformatician, so the scientists who could be designing the next experiment end up waiting for someone else to run the last one.

The solution: Claude as a bioinformatics agent

Biomni built on Claude to run validated bioinformatics pipelines and deliver annotated, reproducible reports, with built-in biosafety controls and full methodology documentation.

• 800 times faster bioinformatics analysis, 35 minutes instead of three weeks
• Cloning experiment designs validated as equivalent to a 5+ year expert in blind testing
• Claude connected to 150 tools, 59 databases, and 106 software packages

Literature synthesis

FutureHouse (opens in new tab) built specialized research agents on Claude to help scientists stay current across a literature base that doubles every few years.

The challenge: science outpacing scientists

Biomedical literature is growing faster than any researcher can track. Staying current across relevant studies, synthesizing findings, and identifying novel directions can take months — time that isn't being spent on the research itself.

The solution: specialized agents for scientific discovery

FutureHouse built four Claude-powered agents spanning literature search, analysis, novelty assessment, and drug discovery. Rather than replacing researchers, the agents compress the front end of the scientific process — surfacing what's known, flagging what's new, and freeing scientists to focus on what's next.

• Literature reviews completed in days instead of months
• Four specialized agents covering the full research workflow, from literature search to drug discovery

Regulatory and clinical development: from data to submission

While discovery teams accelerate the front of the pipeline, regulatory and clinical development teams use Claude to clear the documentation work that stands between a finished trial and patient access.

Regulatory documentation: Novo Nordisk

Novo Nordisk (opens in new tab) tackled the documentation bottleneck that delays treatments from reaching patients.

The challenge: documentation delays blocking patient access

Each new treatment requires mountains of documentation: clinical study reports running hundreds of pages, technical device verification protocols, and patient guides written in plain language. Producing a single clinical study report was a multi-month effort, with staff writers averaging only 2.3 reports a year. Each day of delay in bringing a medicine to market can cost up to 15 million dollars in potential revenue, and patients keep waiting.

The solution: the NovoScribe documentation platform

Novo Nordisk built NovoScribe, a generative AI platform on Amazon Bedrock and MongoDB Atlas with Claude as the frontier intelligence, developed using Claude Code. It combines retrieval-augmented generation with expert-approved text and case-specific variables to produce accurate, compliant documentation, starting with clinical study reports and expanding to device protocols and patient materials.

• 10+ weeks to 10 minutes for clinical study documentation
• 95 percent reduction in resources for device verification protocols, from entire departments to single users
• 50 percent fewer review cycles through improved clinical accuracy
• Complete study booklets produced in under one minute
• An 11-person team that stays agile while expanding capabilities

Clinical workflows: Bluenote

Bluenote (opens in new tab) built AI agents that let life sciences researchers spend their time on science, not paperwork.

The challenge: documentation that consumes researchers

Bringing a new treatment to market demands as much paperwork as it does research. Regulatory submissions, clinical trial reports, and quality validations are essential — but they consume days or weeks of a researcher's time.

The solution: Claude-powered agents for clinical and regulatory workflows

Bluenote builds AI agents that automate the documentation work woven throughout clinical operations—regulatory submissions, study reports, validation protocols, compliance forms. Their agents process regulatory documents, generate technical reports with proper citations, and handle complex multi-step workflows, all with robust guardrails, data traceability, and highlighted calls to action for human experts to contribute additional context and review. Claude serves as the default model for scientific and technical documentation, chosen for its citation capabilities and accuracy in a domain where every claim must trace back to a source.

• 50–75% faster regulatory document production
• 10x faster protocol analysis for scientists
• Multi-hundred-page scientific documents with tables, figures, and citations generated in minutes
• Compliance gaps flagged automatically against the latest regulatory guidelines