Recommended Free Tools
To observe an AI agent in Power BI, first capture its OpenTelemetry data, send it to Azure Monitor Application Insights, and use Log Analytics to select and aggregate the results before loading them into Power BI. Application Insights and Log Analytics are where you inspect and query telemetry; Power BI is the downstream reporting layer, not the agent tracing SDK or trace backend.
What the end-to-end pipeline does
The documented Microsoft-centered route has four parts: instrumentation at the agent boundary, telemetry collection in Application Insights, querying in Log Analytics, and reporting selected query results in Power BI. Each layer answers a different question:
As an Amazon Associate I earn from qualifying purchases.
- OpenTelemetry instrumentation: records operations and other signals from an agent or the application that calls it.
- Application Insights: receives and stores telemetry so you can explore and troubleshoot it in Azure Monitor.
- Log Analytics: lets you query the stored data with KQL and shape it for reporting.
- Power BI: turns selected query results into a dataset, dataflow, and report, depending on the route you choose.
This is a representative Microsoft/Azure implementation, not a universal configuration for every language, agent framework, hosting provider, or Power BI tenant. Instrumentation, available fields, permissions, licensing, and ingestion support vary by environment.
Where should tracing begin?
For an agent hosted in Microsoft Foundry
Connect an Application Insights resource to the Foundry project, then start with server-side tracing. Foundry stores traces using OpenTelemetry semantic conventions. This is the lowest-effort starting boundary when you need visibility into operations handled by the hosted agent service.
#1 Best Overall
Add client-side instrumentation when you also need spans from your own application code around agent calls—for example, work performed before a request is sent or after a response returns. Server-side and client-side tracing answer different questions: the hosted-service view shows what happens within that boundary, while application instrumentation can expose custom logic outside it. Instrument only the boundaries you need, and check that their operations can be interpreted together in your telemetry.
For code running outside the hosted service
Choose instrumentation for the actual framework and language. Microsoft documents Python integrations using microsoft-opentelemetry and use_microsoft_opentelemetry, with Azure Monitor export and options for agent identity. The code samples in that framework guide require Python 3.10 or later. Its approach is for the integrations it documents, not a universal setup to paste into every Python agent.
Rank #2
For the OpenAI Agents SDK, Microsoft’s example uses the OpenTelemetry SDK and provider with the Azure Monitor exporter. When an Application Insights connection string is absent, the example uses console export for local debugging. Other frameworks may expose different setup APIs or capture controls; follow the integration documented for the framework you actually run.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11How do you configure export and agent identity?
Configure the exporter to send data to the intended Application Insights resource. Microsoft’s examples use an Application Insights connection string and can attach agent ID or name attributes. Keep connection strings and other secrets in deployment environment configuration or an appropriate secret store rather than embedding them in shared source code.
- Identify the destination. Confirm which Application Insights resource and workspace should receive the agent’s telemetry, especially when development, test, and production environments are separate.
- Configure the integration. Follow the framework-specific Microsoft setup for its exporter, connection string, and identity options. Do not assume configuration names or defaults are identical across integrations.
- Check identity in the emitted data. After deployment, verify that the service and agent identity appear as expected in the telemetry. A connection to the right resource alone does not confirm that the dimensions you need for filtering or reporting are present.
Should telemetry include prompts, responses, or tool content?
Decide on content capture before enabling it. Depending on the integration and its settings, telemetry may include prompts, responses, tool arguments, or tool results. That content can contain personal, confidential, or otherwise sensitive information. The documentation exposes content-capture settings, including experimental GenAI convention options for some integrations, but the switch and behavior are not the same across all packages.
- Review what the chosen instrumentation captures and where that data will be stored.
- Check who can access the Application Insights resource and Log Analytics workspace, along with retention requirements and policies.
- Enable content capture only when the troubleshooting or analysis need justifies collecting it.
- Test with representative data and inspect emitted telemetry to confirm what is actually present.
Do not infer that a setting from one integration controls another. Confirm the specific package’s current behavior and configuration.
Rank #4
How do you verify that traces arrive?
- Run a representative agent interaction after connecting the destination and configuring any external instrumentation.
- Open the relevant Application Insights or Azure Monitor experience for the connected resource and inspect the trace data.
- Check for the root agent operation and the child operations that matter to your workflow, such as relevant calls or custom application work.
- Confirm the service and agent identity, timestamps, and other dimensions you intend to use in queries are present.
- Resolve missing or misidentified spans before building a report; a dashboard cannot recover operations that were never captured or exported.
For the documented Foundry framework workflow, Microsoft says traces typically appear within 2–5 minutes after agent execution. That is an expected interval for that setup, not a service-level guarantee. If data does not appear, first verify that the agent ran, that the intended Application Insights resource is connected, and that the integration and exporter are configured for the code being observed. Then inspect the available telemetry and schema in the target workspace.
Which Log Analytics data should feed a report?
Application Insights represents common OpenTelemetry signals in tables including AppTraces for logs, AppRequests for incoming requests, AppDependencies for outgoing dependencies, AppExceptions for exceptions, and AppMetrics for metrics. The signal, instrumentation, and workspace affect which tables and attributes are available. Check the schema in the actual workspace rather than assuming every field shown in an example exists for your agent.
Use KQL to narrow data to the service, reporting period, and dimensions relevant to the question, then aggregate it before loading it into Power BI. For example, an operational report might summarize agent operations over time and separate results by a verified service or agent dimension. A troubleshooting view may need individual operations, while a trend dashboard usually benefits from summarized results. Build the query against fields confirmed in the workspace, and validate both its results and its time range before export.
Keep detailed trace investigation in Azure Monitor and Log Analytics. Send Power BI the selected, report-ready data rather than treating a BI dataset as a replacement for trace exploration.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How do you choose a Power BI loading route?
| Route | Best fit | Where preparation happens | Key consideration |
|---|---|---|---|
| Power BI (as an M query) | Bringing a validated Log Analytics query into Power BI Desktop. | The Log Analytics query selects and shapes results; Power BI can continue modeling them. | Query export is subject to Log Analytics Query API limits and may return partial results when a response exceeds the maximum size. |
| Power BI (new Dataset) | Creating a dataset directly in the Power BI service from the query results. | The Log Analytics query prepares the selected results; modeling and reporting continue in the Power BI service. | Creating the dataset directly requires workspace write permission; Microsoft gives Log Analytics Contributor as an example. |
| Azure Log Analytics connector in a Power BI dataflow | Reusable preparation or transformations that should be shared before reports consume the data. | Preparation is handled in the dataflow, which can serve as a shared upstream transformation layer. | Dataflows and related advanced capabilities may require Pro or Premium, depending on the feature and tenant. |
In Log Analytics, run and validate the KQL query before using the export options. Choose Power BI (as an M query) for the Desktop route or Power BI (new Dataset) to create a dataset in the service. A dataflow uses the Azure Log Analytics connector when shared preparation is useful. These choices differ in where transformations and modeling occur; select based on reuse, reporting workflow, and the tenant’s access and licensing arrangements.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →What permissions, refresh, and export limits should you plan for?
- Export to Power BI Desktop: requires workspace query-read permissions. Microsoft names Log Analytics Reader as an example.
- Create a dataset in the Power BI service: requires workspace write permission. Microsoft names Log Analytics Contributor as an example.
- Incremental refresh: requires a datetime field in the query output.
- Advanced Power BI capabilities: sharing, scheduled refresh, dataflows, or incremental refresh may require Pro or Premium depending on the feature. Check the current licensing and permissions in your tenant before implementation.
- Query response size: Log Analytics Query API limits can cause an export to contain partial results when the response exceeds the maximum size. Filter and aggregate to a manageable result set, and verify that exported results are complete enough for the report.
How should you handle preview ingestion paths?
The cited Azure Monitor OpenTelemetry collection documentation is marked Preview. It describes collector export over OTLP HTTP/protobuf and notes other limitations for particular ingestion paths. Preview terms, supported protocols, and constraints can change. Verify the current Azure Monitor documentation and the capabilities of your target environment before using that collector path for production workloads.
Quick Recap
What to check when the report is empty or incomplete
- No traces in the monitoring resource: confirm the agent ran, the correct resource is connected, and the framework-specific exporter is configured and enabled.
- Some expected operations are missing: check whether they fall outside the tracing boundary. Add client-side instrumentation for custom code that server-side tracing does not cover.
- Queries return no rows or missing dimensions: verify the workspace’s actual table and field schema, then adjust the KQL to the signal and attributes that are present.
- Power BI results are smaller than expected: inspect the query time range and filters, reduce the response with appropriate aggregation, and check for Query API export limits that may produce partial results.
- Export or refresh is denied: confirm the user has the required workspace query-read or write permission for the selected route, as well as the relevant Power BI capability and tenant licensing.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
