Agent Observability: No Span for Agent Invocation and unclear span naming

[m-gheini]

• Package Name: azure-ai-projects
• Package Version: 2.0.0b2
• Operating System:
• Python Version:

Describe the bug
Currently, the auto-generated spans in Application Insights lack clear, descriptive names and are missing critical agent invocation spans. This makes it difficult to:

1. Follow execution flow: Spans are not clearly labeled (e.g., "AI unknown" instead of agent name or operation)
2. Track agent invocations: No top-level "invoke_agent" span to identify when an agent starts processing
3. Identify tool calls: Tool execution spans don't clearly show which tool was called (e.g., "execute_tool" vs "execute_tool <tool_name>")

To Reproduce
Steps to reproduce the behavior:

1. Run an agent and setup observability using configure_azure_monitore.

Expected behavior
Connected spans with clear names and spans for agent invocation.

Screenshots

[github-actions]

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @dargilco @glharper @howieleung @kingernupur @nick863 @trangevi @trrwilson.

[github-actions]

Hi @m-gheini, I'm an AI Support assistant here to help with your issue. While the Azure SDK team reviews your request, I wanted to provide some possible tips and documentation that might help you in the meantime.

• Verify that your observability configuration via the SDK’s public API (using configure_azure_monitore) is correctly applied when initializing and invoking your agent. See the Azure SDK logging documentation for guidance on configuring telemetry.
• Ensure that explicit names for agents and tools are supplied where supported. This can help generate more descriptive spans rather than defaulting to generic labels.
• Review your agent invocation code to confirm that it triggers the expected telemetry context. Adding logging around the invocation may help verify that a span is created for the top-level "invoke_agent" operation.
• Compare your implementation with available sample code to ensure that your setup for agent operations and telemetry is aligned with common usage patterns.

The Azure SDK team will respond to your issue shortly. I hope these suggestions are helpful in the meantime. If this comment helped you, please give it a 👍. If the suggestion was not helpful or incorrect, please give it a 👎. Your feedback helps us improve!

[dargilco]

Thank you @m-gheini for creating this issue! The next beta update will include tracing improvements. I'll let @M-Hietala reply to your comments, but he (like many other folks here) are on vacation until the New Year.