Bloomberg Debuts Enterprise MCP to Fuel Financial AI Agents

Most generative AI tools struggle with the "data readiness" problem: they can retrieve a price, but they lack the context to understand its currency, calculation method, or relevance. The new Enterprise MCP platform addresses this by pairing Bloomberg’s massive dataset—covering over 100 million securities—with workflow-focused Skills. This allows AI agents to interpret results correctly, reducing the need for manual reconciliation or identifier hunting.

Tony McManus, Global Head of Enterprise Data and Indices at Bloomberg, noted that the bottleneck for financial firms has shifted from model capability to the data itself. Without clear indicators of what a specific value represents, agents often fail to answer basic questions about positions or risk. The new protocol provides a standardized interface that allows agents to find the right information using natural language, effectively automating tasks like identifying outliers, reviewing trade deviations, and assessing sanctions exposure.

While Bloomberg manages the hosting and governance of the protocol, firms retain full control over their own agents, models, and prompts. The initial release integrates with Bloomberg's Data License Plus, with plans to incorporate real-time data for intraday monitoring and pre-trade checks in the near future. This development marks a significant move by Bloomberg to shape the open standards of agentic AI, as the company currently leads the MCP Financial Services Interest Group.