Catalyst Copilot is an AI-powered agent that Amgen embedded into its internal knowledge management platform, Catalyst, to make R&D knowledge easier to find and use. Amgen’s R&D teams generate and rely on thousands of digital assets—PowerPoints, PDFs, trial reports, articles, and SharePoint folders. Drug developers often had to search the intranet, click through long lists of links, and manually open and review documents. That process could take hours or even days, slowing down knowledge transfer and, ultimately, drug development. The knowledge management team wanted a more intuitive, “ask-a-colleague” experience. Using Microsoft Copilot Studio—a low-code platform for building intelligent agents—they created Catalyst Copilot in about six weeks. The agent ingests Amgen’s carefully curated R&D resources and uses a natural language Q&A interface so researchers can simply ask questions and get synthesized answers, instead of hunting for files. In short, Amgen built Catalyst Copilot to: - Reduce the time it takes to find relevant R&D information and experts. - Turn a scattered digital knowledge base into an accessible, conversational experience. - Support the broader goal of using AI to enable better science at scale and help move potential new medicines forward more efficiently.

Catalyst Copilot helps Amgen’s R&D teams by acting as an intelligent front door to the company’s curated scientific knowledge. Here’s how it works in practice: 1. Natural language Q&A interface Researchers and other R&D roles can ask questions in plain language—for example, about past trial designs, specific molecules, or prior lessons learned. Instead of returning a long list of links, Catalyst Copilot provides a synthesized answer based on Amgen’s existing documents and resources. 2. Reasoning over multiple fact sets The agent doesn’t just keyword search. It can ingest, filter, and reason over large volumes of information, drawing inferences from multiple documents. This helps teams reach faster, more accurate conclusions and reduces the risk of missing relevant prior work. 3. Shortening the discovery cycle By quickly surfacing what the organization already knows and who might have relevant expertise, Catalyst Copilot helps: - Cut down the “white space” time spent searching for information and people. - Remove barriers such as not knowing the exact right question or keyword. - Let teams move more directly to scientific decision-making. 4. Continuous improvement of knowledge curation The Q&A style chat also gives the knowledge management team insight into user intent and common questions. They can use this to refine how content is curated and organized, further improving the quality and relevance of answers over time. Overall, Catalyst Copilot supports Amgen’s goal of using AI to automate parts of the scientific process, distill complex data into usable insights, and help R&D teams progress potential new medicines more quickly and confidently.

Amgen chose Microsoft Copilot Studio because it aligned with both their technology stack and their strategic goals for AI in R&D. Key reasons for choosing Copilot Studio: 1. Existing Microsoft ecosystem Amgen was already an early adopter of Microsoft 365 Copilot and viewed Microsoft as a trusted technology collaborator. Copilot Studio was a natural extension, allowing the team to build on familiar tools and infrastructure. 2. Low-code platform with room to grow Copilot Studio provides a low-code and no-code environment for rapidly building intelligent agents, while still allowing more complex configurations when needed. This helped Amgen’s technology and knowledge management teams: - Build and deploy Catalyst Copilot in about six weeks. - Iterate quickly as they learned from real-world usage. 3. Common platform for agentic workflows By standardizing on Copilot Studio, Amgen can create a common platform for multiple agents across the business. Leadership envisions a future where employees have a single user experience that routes them to the right agent for their needs, whether in R&D or other functions. How this fits into Amgen’s broader strategy: - AI as an information capability: Amgen views drug development as increasingly dependent on how effectively teams can find, parse, and process information. AI is seen as a way to automate parts of the scientific process and distill data into deeper insights. - Better science at scale: One of Amgen’s key metrics for AI success is enabling better science at scale—helping more researchers access high-quality information and insights quickly. - Expanding use of custom agents: Having seen the benefits of Catalyst Copilot in R&D, Amgen plans to explore more agentic workflows in other areas of the business, reimagining how teams learn, collaborate, and make decisions. In essence, Copilot Studio gives Amgen a flexible, scalable way to embed AI into its knowledge management practices and to keep evolving how the organization learns and innovates over time.