Merck and Mastercard show agentic AI depends on enterprise plumbing

This is a useful adoption story because it is not another vendor launch. VentureBeat reports that Merck is getting measurable early results from agentic AI after investing in shared infrastructure: one AI-assisted discovery cycle was cut by a third, regulated marketing drafts are reportedly 99% compliant in first form, and delivery can move 70%-80% faster. Merck's platform team supports 2,500 AWS accounts, multiple Azure subscriptions, Google Cloud integrations, 47 edge locations, and many petabytes of structured and unstructured data, which explains why the story centers on context, governance, MCP, A2A, and agent registration rather than prompts alone. Mastercard's chief data officer also describes agentic experimentation around chargeback and fraud-dispute workflows, where deterministic rules, probabilistic judgments, cost, and reputational risk collide. The takeaway is practical: enterprise agents need plumbing, confidence checks, and risk design before they deliver real savings.

Key details: Merck, Mastercard, May 27, 2026, 33% shorter drug discovery cycle, 70%-80% faster marketing delivery, 99% compliant marketing drafts, 2,500 AWS accounts, 47 edge locations.

Continue swiping for more AI Brief stories.