Bloomberg unveils the Bloomberg Enterprise Model Context Protocol (MCP), an AI access layer for DL+ that helps AI agents discover, understand, and retrieve Bloomberg data across hundreds of millions of securities and thousands of fields through a standardized MCP interface.

MCP combines AI-ready metadata, semantic context, and workflow-focused Skills to interpret data correctly and support production-ready workflows in research, portfolio, risk, and operations.

Key features include AI-ready metadata describing field meaning and calculation, semantic interfaces for natural-language field discovery, broad content access across DL+ Live and Qube Datastore, and task-specific discovery tools for securities and fields.

The MCP approach enables natural-language requests to yield context-rich data, reducing manual lookups and enabling faster, auditable decisions with information such as currency, pricing date, and data type.

Workflow Skills provide reusable procedures (e.g., point-in-time universe retrieval, corporate action adjustments, revision history) with initial skills to identify outliers, review trades against price deviation thresholds, and assess sanctions exposure, with more to come.

Client-controlled execution model where Bloomberg hosts the MCP while clients manage their AI agents, models, prompts, and steering context.

Entitlements and audit controls ensure data is returned only for appropriately licensed requests, with requests resolved against an Operational Datastore and capped by tool-level limits.

Future plans include extending MCP with real-time data for intraday monitoring, pre-trade checks, risk refreshes, and exception handling, and delivering a Terminal subscribers’ version with governed, attributable access via ASKB (Terminal’s AI interface).

This system enables access across DL+ Live and Qube Datastore, supporting broad data availability while maintaining governance and attribution.