Get an AI agent that runs in your own Azure AI project. Several of your projects call the same agent, and it is maintained in one place. Its own identity, access rights and content filters belong to the agent.

In use

The same pattern runs in this practice’s own Azure AI project, where an agent researches every enquiry that arrives through this website’s contact form.

The Agentic Part

The agent runs in Microsoft Foundry, Microsoft’s platform for agents, inside your own Azure AI project. Foundry operates it without code or infrastructure of your own, and we set instructions, model and tools together with you. From several projects a program calls the agent like a service. The agent works with its own identity in Microsoft Entra, with a private network connection, role-based access and content filters, and every version is saved automatically.

Main Features

Your agent runs in your own Azure AI project
Several of your projects call the same agent
A dedicated identity in Microsoft Entra for the agent
Private network connection and role-based access
Content filters per agent
Every version is saved automatically, and a rollback to an earlier version is possible
Publishing to Microsoft Teams and Microsoft 365 Copilot is possible

Use Cases

Several applications in the company need the same AI capability. All of them call the one agent in your Azure AI project.
A team should use the agent in Microsoft Teams. It is published there, with its own identity and its own access rights.
A new version behaves differently than expected. You go back to the earlier version.