Bloomberg launches Enterprise MCP for agentic finance data

Bloomberg on Tuesday launched Bloomberg Enterprise Model Context Protocol, an AI access layer for Data License Plus that lets client AI agents discover, interpret and retrieve licensed data across more than 100 million securities and 50,000-plus fields through a standardized MCP interface.

The product pairs semantic metadata — currency, price type, as-of date, calculation method — with workflow “Skills” for tasks such as outlier checks, trade-deviation reviews and sanctions screening. Bloomberg hosts the MCP server; clients keep control of their own models, prompts and agents. Entitlements are checked before any data returns, and tool-level caps limit how many securities or fields a single call can pull. Tony McManus, global head of enterprise data and indices, said the bottleneck in finance AI has shifted from model smarts to data readiness.

Roadmap items include real-time feeds for intraday risk and pre-trade checks, plus a Terminal-side path that mirrors ASKB conversational access inside desktop workflows. For buy-side and sell-side firms in Toronto and New York already piloting agents, Enterprise MCP is less a chatbot and more plumbing: a way to move generative AI from sandbox demos into audited production without inventing a private data dictionary for every team.