Microsoft is accelerating its agentic ERP, Dynamics 365, by converting its Model Context Protocol (MCP) server from a static toolset into a dynamic, real-time gateway of several ERP functions, while also extending MCP into analytics. All new Dynamics 365 ERP agents will adopt MCP, and existing ones will migrate by December 2025. According to its November 11 product update, this will better position MCP for adaptive, AI-enabled operations across finance, supply chain, HR, and project workflows.
Per the update, the first MCP server introduced at Microsoft Build 2025 offered 13 curated tools and provided a standardized, governed way for AI agents and applications to access ERP data and actions. The newly released dynamic server, now in public preview, replaces that fixed catalog with a flexible framework that allows agents to navigate ERP forms, update fields, and execute available actions through server APIs. This opens access to hundreds of thousands of ERP functions and tens of thousands of forms, all governed by existing permissions, auditing, and security. Independent software vendor (ISV) extensions and tenant-specific customizations are automatically surfaced through MCP, enabling agents to work with environment-specific extensions without new code.
MCP Extends to Analytics
A second server, entering public preview in December, applies the same model-context approach to ERP analytics via Business Performance Analytics. Agents and business intelligence (BI) tools reportedly gain governed access to measures, dimensions, reports, and semantic models. This ensures that operational actions and analytics share the same definitions for metrics like margin or cash flow, improving trust and reducing discrepancies. It also enables natural-language analysis and forecasting through Copilot and Microsoft Teams.
The news update emphasizes the MCP evolution offers IT a unified governance and standardized integration layer that reduces maintenance and accelerates solution delivery. For business leaders, connecting insight to action helps shift teams from retrospective reporting toward proactive performance management.
Partners are already demonstrating real-world value. Examples include RSM’s shop-floor agent, HSO’s payment inquiry automation, Fellowmind’s inbound load creation, Cegeka’s recall impact analysis, Crowe’s unstructured process automation, KPMG’s supplier performance insight, Annata’s service chain coordination, and SignUp Software’s production recovery planning.
Microsoft will share more at Microsoft Ignite in November and Convergence in December, as it continues building MCP into the strategic foundation of adaptive ERP systems that continuously adapt to business needs and AI-driven operations.
What This Means for ERP Insiders
CIOs and IT leaders are getting a cleaner path to scalable automation. The shift to a dynamic MCP server removes one of the biggest barriers IT teams face: one-off integrations that slow new automation projects. With hundreds of thousands of ERP functions now accessible through governed, real-time APIs, CIOs gain a single pattern for building and managing agents across Dynamics 365.
Analytics is about to get less fuzzy, changing decision-making. The analytics-focused MCP server creates something finance teams have long sought: analytics and operations speaking the same language. Because AI agents and BI tools now use shared semantic models from Business Performance Analytics, metrics like revenue, margin, and cash flow no longer shift between systems. Additionally, forecasts and variance alerts can trigger workflow steps directly in the ERP.
Agentic workflows will change daily work. The outlet highlighted early case studies that preview the day-to-day shift ahead and show where value appears first. Essentially, MCP-aligned agents thrive where processes are repetitive, high-touch, or time-sensitive. When adopting similar technology, end users should focus on scenarios where agents create measurable relief and faster throughput, such as when employees navigate multiple screens, perform manual validations, or chase data across systems.
