Merck uses AI to speed up drug research and marketing

Merck is significantly accelerating its drug discovery and marketing processes through the strategic implementation of AI agents. The company has observed a reduction of up to one-third in drug discovery cycles, potentially bringing life-saving therapies to patients a year sooner. In its regulated marketing operations, AI agents are generating compliant materials with 99% accuracy, drastically cutting review times from months to mere days and speeding up delivery by 70% to 80%.

According to Sean Finnerty, VP of Digital Platforms, the success of these agentic AI applications is critically dependent on establishing a strong foundational infrastructure, or 'plumbing.' This approach, learned from early cloud adoption, ensures that thousands of AI agents can be securely managed, registered, and connected to the necessary data and tools.

Merck's infrastructure supports extensive cloud environments and petabytes of data across various repositories. The focus is on building scaffolding for seamless and secure context delivery to AI agents, enabling them to operate efficiently across different platforms like AWS, Azure, and GCP. This integrated approach allows for flexible deployment of AI workloads.

Beyond accelerating research and marketing, AI agents are also being employed for app modernization. They can document architectures, APIs, and network paths, perform security checks, and even write deployment code. This capability contrasts sharply with previous weeks- or months-long application update cycles that were considerably more costly. Challenges like AI hallucination are being managed through engineered guardrails and multi-AI assessments.

Similar to Merck's advancements, Mastercard is exploring AI agents for complex transaction and dispute workflows. These systems aim to streamline labor-intensive processes that involve both structured and unstructured data, improving efficiency while maintaining customer trust. Both companies highlight the need for careful risk assessment and cost-benefit analysis when integrating AI into sensitive operations.