gemini generated image 86c7ay86c7ay86c7

AI Resources for Custom Copilot Extensions

Summary

The 2025 Release Wave 2 of Microsoft Dynamics 365 Business Central introduces AI Resources for Custom Copilot Extensions, a feature designed to make building and running AI-powered tools inside Business Central easier and more affordable.
Developers can now use Microsoft-managed AI infrastructure directly, without maintaining their own Azure OpenAI accounts. For organizations with specific model or region requirements, bringing their own Azure subscription remains supported.
This update simplifies setup, enhances reliability, and unifies billing, empowering both developers and customers to focus on value instead of infrastructure.


Why This Update Matters

Business Central’s Copilot features are growing from convenient add-ons into a foundational part of how the product works.
In earlier releases, ISVs and partners who wanted to build their own Copilot-powered extensions, such as invoice summarization or smart order assistants, had to provision their own Azure OpenAI resources. That meant managing authentication, scaling, security, and separate billing.

With the 2025 release, Microsoft has made AI a native service within Business Central.
Now, developers can tap into the same AI backbone that powers Microsoft’s own Copilot experiences, without the setup hassle.

For customers, this means faster deployment, lower overhead, and consistent data protection aligned with Microsoft’s Responsible AI standards.


Business Value in Practice

This feature represents more than convenience. It directly impacts productivity for developers, partners, and customers alike.

  1. Reduced operational overhead
    There’s no need to manage servers, API keys, or performance scaling. Microsoft handles this globally.
  2. Simplified onboarding
    Developers can connect to AI capabilities directly from Business Central’s developer toolkit without configuring Azure resources.
  3. Improved reliability
    Load balancing, throttling, and uptime management are all handled by Microsoft’s infrastructure.
  4. Built-in security and compliance
    Data governance, regional residency, and harmful-content safeguards are applied automatically.
  5. Unified billing
    Instead of multiple invoices from different vendors, customers receive one combined AI usage statement through Business Central.

This unified approach gives both small partners and enterprise developers a straightforward way to create scalable, AI-powered solutions.


Real-World Example

Consider an ISV that builds a “Service Order Analyzer” extension.
The extension reads support tickets, summarizes issues, and recommends next steps for the service team.

Before Wave 2:

  • The ISV had to manage an Azure OpenAI instance, deploy its own GPT models, and monitor usage manually.
  • Customers had to reconcile separate AI usage bills.

Now:

  • The ISV connects to Microsoft’s managed AI resources through Business Central’s toolkit.
  • The AI automatically scales, usage is tracked centrally, and costs are added to the customer’s Business Central invoice.

This streamlines both development and support, while maintaining predictable billing.


Developer Guide: How It Works

Business Central now includes a new codeunit called System.AI.AzureOpenAI, which provides two authentication options.

Option 1: Use Microsoft-managed AI resources (recommended)

AzureOpenAI.SetManagedResourceAuthorization(
    Enum::"AOAI Model Type"::"Chat Completions",
    AzureOpenAIAccountName,
    AzureOpenAIApiKey,
    AOAIDeployments.GetGPT4oLatest()
);

What it does:
This method connects your extension to Microsoft’s built-in AI environment. Business Central automatically handles authentication, scaling, reliability, and billing. Developers don’t need to configure an Azure endpoint.


Option 2: Use your own Azure OpenAI subscription (advanced)

AzureOpenAI.SetAuthorization(
    Enum::"AOAI Model Type"::"Chat Completions",
    GetEndpoint(),
    GetDeployment(),
    GetApiKey()
);

What it does:
This version connects to a custom Azure OpenAI setup. It’s ideal for ISVs that use specialized or fine-tuned models, have strict data region requirements, or want full control over their own AI costs.


Billing and Cost Structure

Microsoft now supports two AI billing options:
the built-in Business Central AI billing model and the Bring Your Own (BYO) Azure OpenAI subscription model.

AreaMicrosoft-managed AI ResourcesYour Azure OpenAI Subscription
Billing modelCopilot Credits (usage-based)Token-based (Azure OpenAI pricing)
InvoiceIncluded in Business Central billingSeparate Azure invoice
Setup effortAutomaticManual
MaintenanceManaged by MicrosoftManaged by developer
ComplianceMicrosoft Responsible AI standardsDeveloper responsibility
Best forMost users and ISVsPartners needing custom models

Understanding Copilot Credit Billing

Microsoft bills Copilot features through Copilot Credits, a standardized model across Dynamics 365 and Power Platform.

Usage TypeAverage Credits per AI callExample Monthly VolumeEstimated Cost
Light AI (simple text answers)210,000 callsAbout $20
Moderate AI (data lookup or short summaries)1510,000 callsAbout $150
Complex AI (multi-step reasoning, large documents)10010,000 callsAbout $1,000

(1 Copilot Credit ≈ $0.001 USD based on Copilot Studio’s official guidance.)

When you bring your own Azure OpenAI setup, pricing depends on tokens:

  • GPT-4o input ≈ $0.005 per 1,000 tokens
  • GPT-4o output ≈ $0.015 per 1,000 tokens
    A similar workload would cost about $20–25 per month but requires full management.

Partner Monetization and Customer Simplicity

The AI resource model complements, not replaces, AppSource monetization.
ISVs still sell and license their apps through AppSource while AI usage is billed separately inside Business Central.

Customers benefit from:

  • Transparent AI usage tracking
  • Predictable costs
  • No need to reconcile separate invoices across multiple vendors

This structure also helps Microsoft maintain consistency in security, privacy, and compliance across all Business Central tenants.


Availability


Getting Started

  1. Update your sandbox environment to the 2025 Wave 2 version (v27.0).
  2. Access the System.AI.AzureOpenAI codeunit.
  3. Use the managed authorization method for a quick setup.
  4. Experiment with small prompts or Copilot extensions before scaling up.
  5. Monitor AI usage through your Business Central admin center billing summary.

Final Thoughts

The introduction of AI Resources for Custom Copilot Extensions in Business Central 2025 is a key step toward making AI development as easy as app development.
It delivers enterprise-grade reliability and transparent billing, while giving developers freedom to innovate with less operational complexity.

Whether you are an ISV enhancing your AppSource app or a company experimenting with AI automation, this update provides the foundation to build confidently within the Microsoft ecosystem.

Recommended next step:
Try setting up a simple Copilot extension using Microsoft’s managed AI resources in your sandbox environment and explore the new development patterns before the January 2026 general availability release.


References

Leave a Reply

Your email address will not be published. Required fields are marked *