Set up AI features in the Dynamics 365 AI Hub

Completed

In Dynamics 365 Sales, agents are integrated in every key stage of the selling process. This unit walks through how to get started and covers the common prerequisites and cost considerations that apply before you enable any AI feature.

The first place to start is the Dynamics 365 AI Hub.

Explore the Dynamics 365 AI Hub

The Dynamics 365 AI Hub is your central starting point for setting up and managing AI agents in Dynamics 365 customer engagement apps. Think of it as a control panel: it's where admins go to create agents, monitor their performance, and access learning resources.

To access the AI Hub in the Sales Hub app, go to Change area in the lower-left corner, select App Settings, then navigate to General settings > Dynamics 365 AI hub.

The AI Hub includes four main areas:

  • Agent manager: Create and configure AI agents for tasks like lead qualification and opportunity research.
  • AI optimization hub: Monitor agent performance and fine-tune settings using dashboards that show engagement and business impact.
  • Learn how AI agents work: Access videos and documentation to understand how agents function and how to use them effectively.
  • Architecture, security, privacy, and compliance: Review technical details and responsible AI information for your organization.

Tip

If you can't access the AI Hub, it's likely a permissions issue. Your admin may need to assign you the correct security role. Learn more in Access Dynamics 365 AI Hub.

Prerequisites for agents and Copilot

Before you can create an AI agent, there are a few common prerequisites to verify. These apply across Dynamics 365 customer engagement apps and need to be completed once. You check and complete them directly inside the AI Hub.

To verify your prerequisites, open the AI Hub, go to Agent manager > Create and manage agents, and the Prerequisites page opens automatically on your first visit. Work through the following items:

  • Microsoft Copilot Studio capacity: AI agents run on Copilot Studio. You need sufficient message capacity allocated to your environment. Set this up in the Power Platform admin center.
  • Move data across regions: Generative AI features may process data outside your primary region to reach Azure OpenAI Service. You must explicitly accept these terms in the Power Platform admin center under Generative AI features > Move data across regions.
  • AI prompts: Enable AI prompts in the Power Platform admin center so agents can use the predefined and custom prompts that drive their behavior.

After completing those items, turn on two additional settings in the Power Platform admin center:

  • AI insight cards: Notifies sellers about insights surfaced on their opportunity records.
  • AI agents: Enables the configuration and use of AI agents in Dynamics 365 Sales.

Once all prerequisites show as Done, the Create button activates and you're ready to configure your first agent.

Note

Agent-specific prerequisites, such as granting security role permissions for a particular agent, are covered in each agent's setup guide.

Capacity and billing

AI features in Dynamics 365 are powered by Azure OpenAI Service. Microsoft manages the underlying infrastructure, but understanding how consumption translates to cost helps you plan responsibly.

Cost depends on two main factors: the AI model used and the volume of requests generated. A feature that provides a quick summary when a seller opens a record consumes fewer resources than an agent that continuously monitors hundreds of opportunities and sends outreach emails daily. The more active your agents and the more users interacting with Copilot, the higher the consumption.

To estimate expected costs before you roll out AI features broadly, use the Azure pricing calculator. Once you're live, use Microsoft Cost Management and Billing to monitor actual spending, set budget alerts, and download usage reports.

Tip

Start by enabling AI features for a small group of users before a broader rollout. This gives you a baseline for estimating costs at scale.

For detailed pricing, see Azure OpenAI Service pricing.