How to create Multi-tool Agent in AI Foundry UI?

You can attach multiple tools, but only one instance of each knowledge tool type (File Search, Azure AI Search, Bing Search, etc.). If you need multiple indexes with Azure AI Search, use connected agents instead.

In addition, you can use the tool_choice parameter to control behavior: auto (model decides), required (must call a tool), or none (no tools). Also, adding decision rules in your instructions helps, e.g., "Use File Search before Web Search for internal content."

Docs references:

• Tools overview
• Tool best practices

Hi Parul Paul,

If you want to create a multi‑tool agent in Azure AI Foundry using the UI, the main idea is that the agent uses one model but can work with multiple tools to complete different types of tasks.

By default, an agent can generate text. But when you attach tools, it can do more than just answer questions. For example, it can search documents, run code, or retrieve data from external sources. [learn.microsoft.com]

A tool is simply an extra capability that the agent can use during a conversation. When a user sends a message, the model decides whether to call one of the available tools based on the instructions you provided. The tool runs the task, and the result is returned back to the agent so it can give a better response.

To build a multi‑tool agent in Foundry UI, you usually need to do the following.

First, create or open your Foundry project and make sure a model is already deployed. The agent needs an active model deployment to function.

Next, create a new agent from the Agents section in your project.

While creating the agent, you can enable multiple tools from the Foundry tool catalog. Foundry provides built‑in tools that are ready to use after basic configuration, so no custom hosting is required.

Some commonly used tools include Web search for getting real time information File search for retrieving information from uploaded documents Code interpreter for running Python code or doing calculations.

You can enable more than one of these tools for the same agent. Once enabled, the agent automatically chooses which tool to use based on the user’s request and your instructions.

If the agent is not behaving as expected, check the following.

Make sure the required tool connections are configured Confirm the agent identity has permission to access linked resources Verify that the model deployment used by the agent is healthy Check that uploaded files or data sources are correctly indexed.

Multi‑tool agents are useful when you want one agent to handle different tasks like answering questions from documents, doing calculations, and fetching real time data in a single conversation.

Hope this helps clarify how multi‑tool agents work in the Foundry UI. Let me know if you notice any specific error while creating the agent.

Thankyou!