Muokkaa

Jaa


Available tools in Azure AI Agent Service

When you create AI agents, you can specify different tools in order to help ground the model they use, or extend their capabilities. These tools are categorized as knowledge or action tools.

  • Knowledge tools give the agent access to data sources for grounding responses.
  • Action tools enhance the agent's capabilities by allowing it to run various tools at runtime.

Agents can access multiple tools in parallel. These can be both Azure OpenAI-hosted tools like code interpreter and file search, or tools you build, host, and access through function calling.

Tip

You can use the tool_choice parameter to force the use of a specific tool (like file search, code interpreter, or a function) in a particular run.

Knowledge tools

Tool Description Python SDK C# SDK Basic agent setup Standard agent setup
Grounding with Bing Search Allows your agents to incorporate real-time public web data when generating responses. ✔️ ✔️ ✔️ ✔️
File search Augments agents with knowledge from outside its model, such as proprietary product information or documents provided by your users. ✔️ ✔️ File upload only File upload and using blob storage
Azure AI Search Uses an existing Azure AI Search index as a knowledge base. ✔️ ✔️ ✔️

Action tools

Tool Description Python SDK C# SDK Basic agent setup Standard agent setup
Code interpreter Enables agents to write and run Python code in a sandboxed execution environment. ✔️ ✔️ ✔️ ✔️
Function calling Allows you to describe the structure of functions to an agent and then return the functions that need to be called along with their arguments. ✔️ ✔️ ✔️ ✔️
OpenAPI Specification Connect to an external API using an OpenAPI 3.0 specified tool, allowing for scalable interoperability with various applications. ✔️ ✔️ ✔️ ✔️
Azure functions Use Azure functions to leverage the scalability and flexibility of serverless computing. ✔️ ✔️