You might assume that choosing Microsoft business apps means choosing Microsoft’s AI and nothing else. If you run Microsoft 365 or Dynamics 365, Copilot may seem like the only way to bring AI into your daily work.
But that assumption is becoming outdated.
Microsoft is increasingly supporting a “bring your own AI” approach. Organizations can use Microsoft Copilot, bring additional AI models into Microsoft tools such as Copilot Studio, or connect certain third-party AI services to Microsoft 365 and Dynamics 365 data through supported APIs, connectors, Model Context Protocol (MCP) endpoints, and governance controls. These integrations are subject to security, licensing, and administrative configuration requirements.
Think of it as a Car and an Engine
Microsoft Copilot is Microsoft’s product. Think of Copilot as the car and the AI model as the engine.
Microsoft builds the car: the integration with Word, Outlook, Teams, Power Platform, and Dynamics 365, along with the security, compliance, identity, and governance framework.
The engine is increasingly flexible. Organizations now have more options when it comes to the AI models and services that power their experiences.
So “bringing your own AI” can mean two things:
- Choosing from multiple AI models within Microsoft’s ecosystem.
- Connecting external AI tools and agents to Microsoft business data through approved integration methods.
Three Ways to Bring Your Own AI
- Choose from multiple models in Microsoft Copilot
In addition to OpenAI models, Microsoft has expanded support for Anthropic models in selected Copilot experiences, including Copilot Studio and certain Copilot features in Word, Excel, and PowerPoint. Availability depends on licensing, administrator settings, and regional support.
- Use additional models in Copilot Studio and Power Platform
Organizations can build agents and automation that leverage different AI models through Microsoft Copilot Studio, Azure AI services, and the Azure AI model catalog. This enables the use of OpenAI models, Anthropic models, and other supported models while still benefiting from Microsoft’s security, governance, and administration capabilities.
- Connect external AI tools to Microsoft business data
Microsoft is adopting the open Model Context Protocol (MCP), which allows compatible AI agents and assistants to securely access approved business data and operations exposed through Dataverse and supported Dynamics 365 applications. This approach enables organizations to bring trusted external AI tools into existing Microsoft business processes while maintaining administrative oversight.
The exact capabilities available depend on the Dynamics 365 application, the MCP endpoint being used, licensing requirements, and administrative configuration.
OpenAI, Anthropic, and Google: How Each One Fits
AI Provider
Works in Copilot?
Access MS Data?
OpenAI (ChatGPT)
Yes. OpenAI models power many Microsoft Copilot experiences.
Yes, through integrations, APIs, and connectors.
Anthropic (Claude)
Yes, in selected Microsoft Copilot experiences and Copilot Studio scenarios.
Yes, through integrations including MCP-enabled Dataverse and Dynamics 365 scenarios.
Google Gemini
Not currently offered as a native model option within Microsoft Copilot.
Potentially, through supported connectors, APIs, or intermediary integration services.
Governance Still Matters
More AI choice brings a real risk: agent sprawl.
Microsoft’s answer is Agent 365, a control plane designed to help organizations observe, govern, and secure AI agents across the enterprise. Administrators can register, monitor, and govern AI agents and can use Microsoft tools such as Entra ID, Purview, Defender, and Agent 365 to control identity, security, access, and compliance policies. The available controls depend on the type of agent and how it is integrated into the environment.
BYOAI doesn’t have to mean “anything goes.” You can offer choice while maintaining visibility, control, and security.
What to Consider First
- Licensing: External AI tools typically require their own subscriptions and usage-based fees.
- Supported tools and regions: Not every AI tool works with every Microsoft workload, and availability may vary by geography and licensing.
Permissions hygiene: Review permissions before expanding AI usage. In many Microsoft-integrated scenarios, AI agents operate through existing identities, roles, and access controls. Ensuring users and applications have appropriate permissions helps reduce the risk of exposing information more broadly than intended.
The Bottom Line...
Microsoft isn’t asking customers to bet everything on a single AI provider.
Organizations have more AI options than ever before. Depending on licensing and supported integrations, you can use Microsoft Copilot, leverage additional models within Microsoft services, build custom agents in Copilot Studio, or connect approved third-party AI tools to Microsoft 365 and Dynamics 365 data through supported integration frameworks and governance controls.
The question is no longer:
“Should we use Microsoft’s AI?”
The better question is:
“Which AI fits each job, and how do we connect and govern it responsibly?”
At Dyn365Pros, we help organizations get the most from Microsoft Dynamics 365 and the Power Platform. Whether you’re evaluating Copilot, exploring additional AI models, connecting approved AI tools to your business data, or building custom agents that work with Dynamics 365, we can help you choose the right approach, connect it securely, understand the licensing implications, and establish the governance needed for long-term success.
