Unlocking AI-Driven ERP
As AI agents evolve from assistants to autonomous orchestrators, a key question emerges: How will they seamlessly interface with core business systems like ERP without custom coding hurdles? What if your AI could query customer data, generate sales orders, or automate workflows in real-time, all through natural language? Time will reveal the full transformative impact, but Microsoft’s MCP Server for Dynamics 365 Business Central is poised to bridge this gap, turning ERP into a conversational powerhouse. Building on the foundational MCP concepts we explored in our previous article on MCP access for AI agents, let’s dive into this Business Central-specific implementation – now in public preview as part of the 2025 release wave 2.
What is the MCP Server in the Context of Business Central?
The Model Context Protocol (MCP) Server for Business Central is an innovative, open-standard API layer that exposes key ERP entities—such as customers, items, and sales orders—directly to AI agents. Announced in the 2025 release wave 2 plan, it standardizes how applications deliver contextual data to large language models (LLMs), enabling plug-and-play integrations with platforms like Azure OpenAI and Copilot Studio.
Key Features and Capabilities
From the release plan and early previews, here’s what we know about MCP Server’s core mechanics:
- Entity Exposure and Operations: It surfaces standard Business Central APIs (e.g., customersV3, items) as MCP-compliant “tools.” Initially focused on read operations for safety, it supports CRUD (create, read, update, delete) for select entities, with plans for expansion. Agents can invoke these via natural language, e.g., “List overdue invoices for Client X,” pulling live data without manual exports.
- Seamless AI Integrations: Built for Copilot Studio, it auto-syncs tools into agent workflows. For instance, in a demo from Directions EMEA 2025, agents connected via MCP to automate sales order creation from email inquiries. It also works with Azure OpenAI for custom agents, enabling conversational ERP access in Teams or custom apps. Non-compliant agents fall back to basic APIs, but MCP provides richer, contextual responses.
- Real-Time Synchronization and Security: Changes in Business Central (e.g., new fields or permissions) propagate instantly to connected agents. Security leverages Azure AD for scoped access—agents see only permitted data—ensuring compliance with GDPR and other standards.
For deeper technical setup, check Microsoft’s guide on configuring MCP Server, which covers enabling the feature in BC 27.1 environments and mapping API pages to tools.
Timeline and Getting Started
Public preview kicked off on October 31, 2025, with general availability slated for early 2026 (within the October 2025–March 2026 wave). To test it now, spin up a sandbox tenant in Business Central 2025 wave 2, navigate to the MCP Server Configurations page, and link your first API entity. Early adopters report setup in under 30 minutes, with immediate wins in agent prototyping.
The Road Ahead: From Preview to Autonomous Enterprise
While preview limitations like read-only defaults persist, MCP Server signals Microsoft’s push toward an agentic ERP future – where AI not only analyzes but acts on Business Central data autonomously. As one MVP highlighted at Directions EMEA, “This is a huge leap for Agentic AI in Business Central.” Pair it with wave 2’s enhanced Copilot agents for sales and payables, and you’re on track for workflows that evolve with your business.
Ready to experiment? Head to the release plan for full specs, or drop us a line for a custom demo. What AI-ERP scenario are you tackling next?
That’s a really interesting development – it seems like Microsoft is taking a smart approach to integrating AI with traditional business processes. It’s exciting to see ERP systems becoming more accessible and intuitive.