Aligning Business Strategies

Evolving the Dynamics 365 ERP Model Context Protocol (MCP) Server 

https://aka.ms/IgniteMCPBlog

At Microsoft Build 2025, the Dynamics 365 ERP Model Context Protocol (MCP) server was introduced as a foundational step in connecting AI and enterprise resource planning (ERP) systems through a shared, governed protocol. The MCP server provides a standardized way for AI agents, applications, and services to securely access ERP data and execute business actions within Dynamics 365.

That first release represented a milestone. It made ERP data and logic discoverable and callable through a standard set of tools, paving the way for agentic innovation across finance and operations.

Now, Microsoft is extending that foundation with two major advancements:
  • The Dynamics 365 ERP MCP server is evolving from static to dynamic. This unlocks hundreds of thousands of ERP functions for secure, real-time use by agents, developers, and applications. The new dynamic server is now in public preview.
  • A new Dynamics 365 ERP MCP server for analytics is being introduced. It applies the same model-context principles to business intelligence and insights, with public preview of the server coming this December.

Together, these capabilities advance how ERP systems, data, and AI agents connect: simplifying extensibility, supporting accelerated development, and enabling consistent governance across business processes.

The Model Context Protocol: A shared foundation for ERP and AI

The Model Context Protocol (MCP) defines a common language for how agents, applications, and services interact with enterprise data and business logic. MCP provides a unified framework, instead of relying on custom APIs or point-to-point integrations. MCP standardizes access to ERP operations, helping ensure consistency, context, and control.

In practical terms, MCP enables developers to:
  • Build once and reuse across multiple ERP applications and environments.
  • Expose new ERP actions or data endpoints dynamically to the agent framework.
  • Maintain consistent data access, permissions, and auditability across all integrations.

This framework helps simplify how partners and customers create and extend agentic experiences. An agent built to reconcile accounts or process supplier invoices can use the same MCP standards to trigger journal entries, validate transactions, or retrieve KPIs across Finance, Supply Chain, Human Resources, and Project Operations–without custom code or repeated integration work. Adding the MCP server to your agent provides access to data, operations, and analytics in the server for the agent’s security role.

The diagram illustrates the architecture of Microsoft's Copilot Studio, showcasing the integration of MCP servers, Dynamics 365, and various components such as agents, data, and analytics.

AI-generated content may be incorrect.

As organizations move from systems of record to systems of action, MCP provides the connective tissue. It brings ERP data, analytics, and AI together under a single, trusted framework. The evolution from static to dynamic MCP—and the addition of MCP for analytics—marks a significant step in making ERP systems more adaptive, extensible, and aligned with how businesses operate today.

From static to dynamic

The initial ERP MCP server launched at Build 2025 as a static implementation with a curated set of 13 tools for Dynamics 365 Finance and Supply Chain Management. These tools exposed high-value ERP operations through MCP, helping early adopters test and validate real-world agent scenarios.

However, static systems, have natural limits. The set of available tools was fixed, updates required code changes or redeployments, and extending functionality took time. As developers began building more complex agents that span processes and applications, a new level of flexibility became essential.

The new dynamic ERP MCP server, now available in public preview, addresses that need. It transitions from a fixed catalog of tools to a dynamic framework that adapts as business needs evolve. You can now build agents that can work with data and perform nearly any function available to a user through the application interface, without the need of custom code, connectors, or APIs.

How it works

The tools in the new server work by enabling the agent to navigate server forms to complete tasks. Agents work with the application data and business logic through server APIs just like humans perform tasks in the application client. Static tools with very specific actions, like Find Approved Vendors or Release Purchase Requisition Lines, have been replaced. The agent uses the new tools to open forms, set field values, and click actions available on the form.

This evolution unlocks hundreds of thousands of ERP functions across tens of thousands of forms. These forms—in Finance, Supply Chain, Human Resources, and Project Operations—are instantly accessible through MCP. Each function inherits ERP’s existing controls for permissions, auditing, and security, allowing IT to scale innovation while preserving compliance.

The context provided to the agent through the MCP server dynamically updates with each tool call based on the agent’s security permissions and application configuration, extensions, and personalization. This helps ensure the server is working with an accurate view of available actions and data for the given context. Additionally, it automatically makes available in the MCP framework the ISV extensions and customizations in the environment.

Consider an agent that performs supplier selection for purchase requisitions. When submitting the requisition, the agent gathers and analyzes price and supplier performance information from the ERP applications. It then reasons over the data to select the appropriate supplier for the requisition. The agent updates requisition record with the selected supplier by opening the form, populating the appropriate fields and saving the record. The agent then submits the document to an approver for review by selecting the available workflow action on the form. The native capabilities of the MCP server support these interactions between the agent and ERP data and business logic, among many others.

Extending MCP to analytics

The new Dynamics 365 ERP MCP server for analytics extends the same model-context approach to data, metrics, and insights. It provides governed access to the ERP analytics data available in Business Performance Analytics (BPA)—including measures, dimensions, reports, and semantic models. AI agents and analytics tools can reason over consistent, trusted definitions.

This new server creates a bridge between ERP transactions and business intelligence. It ensures that agents and users reference the same definitions for revenue, margin, or cash flow when generating insights or forecasts—reinforcing trust and mitigating discrepancies between analytics systems.

By aligning analytics and operations under one protocol, ERP data becomes agent-ready and explainable. Access AI-driven analysis, forecasting, and variance detection with natural language prompts and surface results directly into the applications people use every day, including Microsoft Copilot and Microsoft Teams.

The MCP server for analytics also creates new opportunities for the partner ecosystem. Developers can build analytics extensions that plug directly into ERP data models while maintaining full governance and auditability.

Why this evolution matters

This announcement is not just about new APIs—it’s about creating a governed, extensible foundation for innovation across ERP.

For IT organizations, the dynamic ERP MCP server simplifies extensibility and accelerates delivery. MCP standardizes how these capabilities are published, managed, and secured—allowing IT to focus on innovation rather than maintenance.

For business leaders, the analytics MCP server connects insight to action. Forecasts, reconciliations, and recommendations produced by AI are grounded in ERP’s trusted definitions, helping to ensure accuracy and supporting compliance. This helps finance and operations teams shift from retrospective reporting to proactive performance steering.

Together, the servers create a single, unified model where data, analytics, and process automation operate in sync. This helps reduce friction, increases visibility, and accelerates the time from signal to action.

By making hundreds of thousands of ERP functions accessible through a dynamic, governed model, and by extending that same model to analytics, Microsoft is building the foundation for ERP systems that evolve continuously with the business—securely, intelligently, and at enterprise scale.

Partner innovation and ecosystem impact

These changes also expand opportunities for Microsoft’s partner ecosystem. ISVs and system integrators can now build and deliver agents faster. Know that MCP provides a consistent integration and governance layer across all ERP workloads.

The dynamic MCP server enables partners to:

  • Create reusable, compliant ERP agents that can be deployed across multiple tenants.
  • Publish agents that tap into hundreds of thousands of ERP functions, including extensions and customizations, without deep code dependencies.
  • Deliver analytics and insight-driven applications powered by the new MCP for analytics.

Early partner adoption is already demonstrating the potential of this architecture. These examples highlight how MCP can reduce integration overhead, standardize access, and help partners deliver new business value faster, while maintaining enterprise-grade trust and compliance.

  • RSM: Shop Floor by RSM empowers manufacturers to operate as frontier firms by minimizing downtime, improving quality. It enables teams to resolve production issues in real time, driving greater resilience and growth. This intelligent agent transforms the shop floor with automation, actionable insights, and seamless collaboration to set a new standard for operational excellence in manufacturing.
  • HSO: The PayFlow Agent enhances invoice payment efficiency in accounts payable by automating payment inquiries and delivering real-time status updates. PayFlow streamlines payment processes, minimizes manual intervention, and accelerates resolutions. Empower finance teams and project managers to manage “pay when paid” terms with transparency. This intelligent automation enhances supplier relationships and ensures vendors receive timely, accurate information about their payments.
  • Fellowmind: From inbox to Inbound Load in seconds. The Inbound Load Agent transforms emailed delivery notes into inbound loads in Dynamics 365, eliminating repetitive data entry, reducing manual error and accelerating and streamlining workflows.
  • Cegeka: The Quality Impact Recall Agent identifies product quality failures and traces their downstream impact across inventory and customers. It orchestrates potential recall efforts, ensuring timely notifications and corrective actions to safeguard safety and brand reputation.
  • Crowe: AP automation is table-stakes today, and ERP automation shouldn’t end there. From sales orders and contracts to customer and vendor interactions, Crowe’s AI Agents handle the unstructured, high-touch processes that traditional automation can’t reach.
  • KPMG: KPMG’s Supplier Performance Insight Agent automates critical finance workflows by integrating internal ERP data with external market signals to give leaders a complete view of supplier relatability and risk. This showcases how AI can make ERP faster, smarter and more connected for organizations.
  • Annata: Annata’s Autonomous AI Agents automate and accelerate complex service processes—turning multi-day breakdowns into minutes of coordinated action. By connecting directly to enterprise data and workflows through the MCP server, the solution secures parts, creates appointments, and orchestrates the entire service chain to maximize equipment uptime. 
  • SignUp Software: The Axtension Downtime Agent steps in when a machine fails in a production environment or a resource becomes unavailable. The Downtime Agent suggests resolutions and generates alternative production plans, factoring in constraints like promised delivery dates and capacity. Multiple simulation scenarios can be validated and published, enabling organizations to maximize production efficiency.

Building the foundation for adaptive ERP

Building on MCP is a strategic priority for Dynamics 365 applications. ERP teams across Microsoft are aligning on a unified approach to help ensure all agents and applications share this common foundation:

  1. All new ERP agents will be built using MCP.
  2. Existing agents will migrate to MCP by December 2025.
  3. All tools hosted on the static ERP MCP server will transition to the new dynamic framework.

This alignment helps ensure that every ERP agent—whether developed internally or by a partner—operates with shared context, consistent governance, and interoperability across Dynamics 365 and the broader Microsoft ecosystem.

What’s next

The dynamic Dynamics 365 ERP MCP server is now in public preview! The MCP server for analytics will enter public preview in December 2025. Over the coming months, Microsoft will continue validating new features with early partners and enterprise customers, gathering feedback to guide the next phase of development. MCP will continue to be a strategic priority in the agent foundation for Dynamics 365 ERP.

Customers and partners interested in joining the preview can contact their Microsoft representative or explore the latest documentation: Use Model Context Protocol for finance and operations apps

Join us in San Francisco, November 18 –21 at Microsoft Ignite for a deep technical dive into Agentic ERP during the Reinvent Finance & Operations with Agentic ERP session. Discover what MCP means for extensibility and learn how to build an ERP roadmap that enables adaptive, intelligent operations.

Looking ahead, we’ll continue the conversation at Convergence in Miami, December 9–11. Explore how AI and MCP are reshaping the future of ERP, secure your spot today.

The post Evolving the Dynamics 365 ERP Model Context Protocol (MCP) Server  appeared first on Microsoft Dynamics 365 Blog.

Leave A Reply

Your email address will not be published.

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More