Creating a Custom Agent in Microsoft Copilot Studio for Business Central

Building intelligent conversational agents that connect to your Business Central data has never been easier. Microsoft Copilot Studio, part of the Power Platform, enables you to create AI-powered agents that can access and present business information in natural, conversational ways. This guide walks you through the process of building a custom agent using Business Central data.

Understanding Microsoft Copilot Studio

Microsoft Copilot Studio is a Power Platform product designed for building conversational experiences and agents connected to various data sources. When working with Business Central, you have multiple options for creating agents:

• Full Copilot Studio agents with complete customization capabilities
• Lightweight experiences available in Microsoft 365 Copilot chat
• Advanced tooling for professional developers

All agents built in Copilot Studio leverage AI foundation models, use tools and knowledge to access data, and present user experiences through an intelligent orchestrator.

Connecting to Business Central Data

There are two primary methods for connecting your Copilot Studio agent to Business Central data:

1. Power Platform Connector: Connects directly to Business Central APIs
2. MCP Server: Uses the Model Context Protocol to connect to Business Central APIs

What is the Model Context Protocol?

The Model Context Protocol (MCP) is an open protocol that enables AI systems to communicate with business applications in a standardized way. This protocol simplifies the connection process, making it remarkably easy to integrate Business Central data into your agents.

Building Your First Agent: A Step-by-Step Example

Let’s walk through creating a laptop research agent that pulls data from Business Central.

Step 1: Configure Business Central

Before building your agent, you need to set up the MCP configuration in Business Central:

1. Navigate to the MCP Configurations page in Business Central
2. Create a new configuration that exposes the Items API
3. Set the configuration to read-only mode for security
4. Activate the configuration

In this example, the Business Central environment contains laptop items categorized using an item category code called “laptop.” These items are set up as non-inventory items representing employee laptop options.

Step 2: Create the Agent in Copilot Studio

Creating the agent is surprisingly straightforward:

1. Open Copilot Studio and click the New Agent button
2. Provide a description of your agent’s purpose
3. Specify the data source and filtering criteria

For the laptop agent, the instructions specify:

• The agent’s purpose: researching laptop options
• The data source: Business Central items
• The filter criteria: item category code equals “laptop”

Copilot Studio‘s built-in AI assistant processes this information and automatically builds the agent framework, suggesting appropriate options and configurations.

Step 3: Add the Business Central Tool

To connect your agent to Business Central data:

1. Navigate to the tools section in your agent configuration
2. Select the Microsoft Business Central MCP server from the Model Context Protocol tab
3. Configure the connection by specifying:
   • Your Business Central environment
   • The MCP configuration you created earlier

That’s all the configuration required. Your agent is now connected to live Business Central data.

Step 4: Test and Publish

You can test your agent directly within Copilot Studio to ensure it’s working correctly. Once satisfied, publish it to your preferred channels:

• Microsoft Teams
• Microsoft 365 Copilot
• Other supported platforms

Using Your Agent in Microsoft 365 Copilot

Once published to Microsoft 365 Copilot, your agent becomes immediately accessible:

1. Open Microsoft 365 Copilot chat
2. Select your laptop agent from the available agents
3. Use predefined prompts or create custom queries

Example Queries

The agent can handle various types of requests:

• Predefined prompts: “Show me available laptops”
• Specific inquiries: “What screen resolutions are available?”
• Custom questions: “Which laptops are best suited for remote work?”

The agent intelligently combines data from Business Central with information from the internet to provide comprehensive responses, including detailed specifications for each laptop option.

Customizing Your Agent

As the agent author, you have full control over the experience:

• Edit predefined prompts to match your organization’s needs
• Add custom prompts for common scenarios
• Adjust the agent’s instructions to refine its behavior
• Configure which Business Central data the agent can access

Licensing and Cost Considerations

Understanding the licensing requirements is essential before deploying Copilot Studio agents:

What’s Included with Business Central

Your Business Central license includes access to certain Power Platform tools, but Copilot Studio requires additional licensing.

Copilot Studio Licensing

To use Copilot Studio, you need to purchase:

• Credit packs for predictable usage
• Pay-as-you-go subscription for flexible consumption

Copilot Studio operates on a usage-based model, which means costs scale with actual consumption rather than user counts.

User Licensing

An important advantage: you don’t need additional paid user licenses for people using your agents. Users consume the credits from your organization’s credit pool, making it cost-effective to share agents broadly across your organization.

Best Practices for Agent Development

When building agents for Business Central, consider these recommendations:

Start simple: Begin with read-only access to a single entity, like the Items table, before expanding to more complex scenarios.

Use clear instructions: The more specific your agent description and instructions, the better the AI can understand and fulfill user requests.

Leverage MCP configurations: Set up multiple MCP configurations for different use cases, each with appropriate security settings.

Test thoroughly: Use Copilot Studio’s testing environment to validate agent responses before publishing to production channels.

Monitor usage: Track credit consumption to understand usage patterns and optimize costs.

Real-World Applications

Beyond laptop research, Business Central agents can serve numerous business functions:

• Customer service: Answering questions about order status, inventory availability, or account information
• Sales support: Providing product information, pricing, and availability to sales teams
• Procurement: Helping employees find and request approved vendors or items
• Financial inquiries: Answering questions about budgets, expenses, or financial reports

Getting Started

Building a custom agent in Microsoft Copilot Studio for Business Central is remarkably accessible, even for users without extensive development experience. The combination of AI-powered agent building, the standardized Model Context Protocol, and Business Central’s robust API infrastructure makes it possible to create powerful conversational experiences in minutes rather than days.

The key is to start with a clear use case, configure your Business Central environment properly, and leverage Copilot Studio’s intuitive interface to bring your agent to life. As you become more comfortable with the platform, you can expand your agents’ capabilities and create increasingly sophisticated experiences for your users.

For organizations already using Business Central, Copilot Studio agents represent a powerful way to make business data more accessible and actionable through natural conversation, ultimately improving productivity and decision-making across the organization.

If you’re planning to scale beyond a simple prototype, it helps to align your agent with the broader Business Central roadmap, security, environments, integrations, and support. Dynamics Folio3 can assist with Business Central implementation and upgrades, Azure-based integration and data migration, managed support/optimization, and deploying Copilot Studio agents that securely access Business Central data.