New AI era in Dynamics 365 Business Central has officially started with the 2025 Wave 2 release, and its numbers across the world are growing, always more (crossed the line of 50.000 online customers with 242 countries and regions supported, so global availability).

The product now has a new logo (aligned with the recently refreshed images of all the Microsoft products), and you should become familiar with it:

The ERP will be enriched in the following months with some important applications features, like for example the new Quality Management module that will support the following:

• quality check for purchased, produced, and assembled items
• possibility to define checks per items, or group of items, or lots/SKU
• possibility to define parameters and attributes to test with pass/fail rates
• support for printing quality certificates and certificates of analysis
• support to put items on quarantine to prevent immediate usage

Also, a new Subcontracting module and many other improvements on reporting, EDI, sustainability, and financials are introduced in this new release, or they will be available soon.

But it’s inevitable that the big emphasis recently is on another theme: AI.

Dynamics 365 Business Central 2025 Wave 2 release (version 27) is extremely focused on introducing AI features, and the same will happen even more in the next waves.

Here is an overview of all the features that make Dynamics 365 Business Central a real “AI-oriented” ERP.

Sales Order Agent and Payable Agents

The Sales Order Agent helps you automate the process of taking sales orders based on customer inquiries about products/items received via email.

The Sales Order Agent monitors a designated mailbox for incoming customer emails regarding item inquiries. When it identifies a potential request, it starts converting the request into an order. For example, it verifies the customer, checks item availability, creates a sales quote, and prepares an email response to the customer with the quote as a PDF attachment.

Some steps require your intervention, such as reviewing e-mail correspondence and assisting the agent as needed. Until an order is created, the agent handles email exchanges with the customer to resolve any missing details and make modifications to the original request as necessary.

Agents work on their own to perform tasks, respond to different events and inputs, reason over the state of data, and fix errors based on user input and the context of the Business Central user experience.

The Sales Order Agent is GA starting from November 2025.

The Payables Agent is a new agent that supports the end-to-end processing of vendor invoices. The Payables Agent monitors mailboxes for incoming vendor invoices, uses AI to analyze invoice content, and shows invoice drafts to agent supervisors for review. Ideally, users make few or no corrections before finalizing drafts, so invoices are ready for approval and posting.

The Payables Agent will be GA from November 2025 in the United States, the United Kingdom, Australia, and New Zealand, with 6 new countries that will be available in the next months (Canada, Denmark, Spain, France, Italy, Germany).

Agent Playground & SDK

The Developer Tools for Copilot in Business Central (or AI Toolkit) will be enriched soon with a set of new tools for supporting Agent runtime and new Copilot capabilities in Business Central (plus supporting also the testability of AI solutions).

The Agent Framework (or Agent Playground) will permit you to create autonomous agents running inside Business Central that can act as a real user (so the possibility to do actions, opening and closing pages, navigating, and more). The best of this framework is that it will permit you to create all these features without using AL code, but everything is in the product.

You will be able to declare your own prompt and tasks for the agent, plus the relative permissions set on data, and when created, it will be executed in the background, exactly like any other standard agents in the product.

The custom agents created with this new framework will support all the features listed below:

The new Agent Playground and SDK will be available in public preview starting from January 2026.

Business Central MCP Server

The Model Context Protocol (MCP) is an open standard developed by Anthropic that enables AI models to securely and efficiently interact with external tools, data sources, and APIs. It acts as a standardized communication layer between AI assistants and the various resources they need to access, such as databases, file systems, web services, and specialized tools.

Key Benefits:

• Standardized Interface: Instead of each AI provider creating their own integration methods, MCP offers a unified protocol that works across different AI platforms and tools, reducing development complexity.
• Simplified tool integration: Developers can easily connect new tools and data sources to AI assistants without modifying the AI model itself. This creates a plug-and-play ecosystem of capabilities.
• Better Performance: MCP enables efficient data streaming and context management, allowing AI models to access real-time information and perform complex multi-step tasks more effectively.
• Extensibility: The protocol supports a wide range of tool types – from simple calculators to complex enterprise systems – making it easy to expand AI capabilities as needs evolve.
• Interoperability: AI assistants can seamlessly switch between different tools and data sources, creating more capable and versatile AI experiences across various platforms and use cases.

Starting from Dynamics 365 Business Central version 27.1, the public preview of the Business Central MCP server will be available:

With the MCP server for Business Central, you can easily connect agents to existing knowledge sources and APIs. You can enable agents to interface directly with Microsoft Dynamics 365 Business Central. Actions and knowledge synchronize automatically. This synchronization facilitates real-time updates and the evolution of functionality. This model simplifies agent development and minimizes ongoing maintenance efforts.

The new Business Central MCP server supports standard and custom API as tools:

When activated, the MCP server will be available to be consumed on tools that support MCP servers, like Copilot Studio (officially the first tool supported by Microsoft for this MCP server).

Business Central MCP Server for Administration

Microsoft has also announced a program to introduce an MCP tool for administrators, available in preview starting from January 2026. Also, this new MCP server will be available for Copilot Studio.

At Directions EMEA in Poznan, I’ve done a session with Duilio about managing multiple Business Central SaaS customers at scale, and in this session, we presented some of the tools we currently use for this task.

One of the tool presented was my D365 BC Admin MCP server (code-named YAMPI), that you can use for free starting from now.

The tool is available on npm as a standalone package that can be installed on GitHub Copilot, Claude, or Cursor.

It can be easily installable on GitHub Copilot via a Visual Studio Code extension available in the VS Code marketplace.

It will permit an easy management of your customers’ tenants at scale, fully using AI and natural language:

More details can be found in the following post.

Conclusion

The integration of AI features into Microsoft Dynamics 365 Business Central represents a transformative leap forward for modern enterprise management. By harnessing intelligent automation, predictive analytics, and adaptive learning capabilities, organizations can unlock unprecedented efficiency, accuracy, and strategic insights. As AI continues to evolve within the Business Central ecosystem, businesses are empowered not only to react to market demands but also to anticipate and shape them, turning data-driven intelligence into a competitive advantage that drives sustainable growth and innovation in an increasingly complex business landscape.

Stay tuned, as you can expect to see more AI features in the coming months. Start adopting them and propose them to your customers.