Catalyst Copilot is an AI-powered agent that Amgen’s R&D knowledge management team built using Microsoft Copilot Studio and embedded into their internal knowledge platform, Catalyst. It was created to address a common challenge in R&D: critical knowledge is spread across thousands of digital assets—PowerPoints, PDFs, reports, SharePoint folders, and more. Previously, a drug developer might spend hours or even days searching the intranet for past trial designs, information on a specific molecule, or lessons learned from earlier work. Catalyst Copilot ingests, filters, and reasons over Amgen’s carefully curated collection of R&D resources. Through a Q&A-style chat interface, it lets researchers ask questions in natural language and quickly surface relevant information and people. The goal is to shorten the discovery cycle by making it easier to find what the organization already knows, so teams can move potential new medicines forward more efficiently.

Amgen built Catalyst Copilot in about six weeks by combining its existing knowledge management work with Microsoft Copilot Studio, a low-code platform for creating intelligent agents and agentic workflows. Key factors that enabled the rapid build: 1. **Strong foundation of curated content** Amgen’s R&D knowledge management team had already done the “hard work” of organizing and curating their knowledge—lessons learned, articles, presentations, and other resources. This meant the AI agent could be layered on top of a well-structured information base rather than raw, unorganized data. 2. **Low-code, extensible platform** Copilot Studio allowed the team to design and deploy the agent with low-code and no-code tools, while still supporting more complex configurations when needed. This reduced development time and made it easier for the knowledge management and technology teams to collaborate. 3. **Existing Microsoft 365 Copilot adoption** As an early adopter of Microsoft 365 Copilot, Amgen already trusted Microsoft as a technology collaborator and had experience with the broader Copilot ecosystem. This familiarity helped them experiment quickly and focus on where AI could have the most impact. The result is Catalyst Copilot, an agent that can analyze, classify, and reason over Amgen’s R&D knowledge, accessible through a natural language Q&A interface inside the Catalyst platform.

Catalyst Copilot is reshaping how Amgen’s R&D teams access and use knowledge by turning information retrieval into a conversational, insight-driven experience. Practical impacts include: 1. **Shorter time to find answers** Instead of starting with “Where is the information I need?”, researchers can start with “What do we already know?” They ask questions in natural language and receive synthesized answers, rather than a long list of links. This reduces the “white space” time spent hunting for documents or the right person to talk to. 2. **Faster, more accurate conclusions** Catalyst Copilot can reason and infer across multiple fact sets, helping teams reach better, more accurate conclusions more quickly. This supports one of Amgen’s key AI success metrics: enabling better science at scale by automating parts of the scientific process and distilling data into deeper insights. 3. **Improved knowledge curation and intent understanding** The Q&A-style chat not only serves users, it also helps the knowledge management team understand user intent and information gaps. They can then refine and more precisely curate content in Catalyst, creating a feedback loop that improves the knowledge base over time. 4. **Accelerated path to potential therapies** By making it easier to find relevant information and connect with the right experts, Catalyst Copilot helps shorten the cycle of discovery. This supports Amgen’s broader mission of getting therapies to patients faster, by treating drug development as an information capability and using AI to streamline how knowledge flows across R&D. Looking ahead, Amgen plans to extend custom agentic workflows built with Copilot Studio into other areas of the business, with a vision of a single user experience that routes employees to the agents they need across the organization.