The financial sector requires large language model (LLM) outputs to be grounded in timely, accurate, and authoritative data due to the high-stakes nature of market movements. This article examines the critical role of the emerging Model Context Protocol (MCP) in meeting these enterprise requirements, specifically within firms like Bloomberg. We explore how MCP acts as the necessary API for the age of agentic AI, facilitating system interoperability, connecting disparate data silos, and enabling the secure, governed deployment of context-aware LLM applications.