Why should you care?

Without observability, CRM copilot degradation shows up only as angry tickets — not as actionable spans.

See it live — copy this example

Paste into an ASP.NET Core 8+ / AgentNest project, then run with dotnet run (set your API keys in user-secrets).

// Program.cs OpenTelemetry
builder.Services.AddOpenTelemetry()
    .WithTracing(t => t.AddSource("AgentNest.Agents").AddAspNetCoreInstrumentation())
    .WithMetrics(m => m.AddMeter("AgentNest.Ai").AddRuntimeInstrumentation())
    .UseAzureMonitor();

What happened?

• OpenTelemetry registers AgentNest agent ActivitySource and custom meter, exporting traces and metrics to Azure Monitor.
• Follow the steps below — typing the code yourself is the fastest way to learn.

Practice next

1. Instrument agents with ActivitySource and Meter("AgentNest.Ai").
2. Create workbook: p95 latency, tokens/min, tool error rate.
3. Correlate App Insights traces with AiAudit rows via operation_Id.
4. Add custom metric rag.hit_rate from RagPipeline.
5. Export same telemetry to Grafana for hybrid cloud tenants.

Remember

OpenTelemetry traces and metrics feed Azure Monitor. Dashboard tokens, tools, RAG, and HTTP together. Link traces to audit records for incident response.

Incident postmortem

ERP agent tool errors spike after plugin deploy.

Outcome: Trace view shows gl.fetch_balance failures; rollback pinpoints bad plugin version.