Skip to main content

​
What are Traces?

As part of your voice-agent stack, traces are used to monitor and observe the behavior of your application in real-time. They provide insights into the execution flow, latency, and performance of your system.

​
Sending Traces to Bluejay

Bluejay supports any traces that conform to the OpenTelemetry standard, this includes OpenInference, Langfuse, OpenLLMetry, and more. To send traces to Bluejay:
  1. Instrument your application code to specify exporting your call traces to Bluejay.
  2. Link your traces to your call evaluations by including the trace_id in your requests to our evaluate endpoint.
  3. Visualize your traces along side your call evaluations in the Bluejay dashboard.

​
Using OpenInference

Send traces to Bluejay using OpenInference by configuring your OpenTelemetry SDK with the appropriate exporter. Below is an example of how to set up the OpenTelemetry SDK to send traces to Bluejay: 
from opentelemetry.sdk import trace as trace_sdk
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk.trace.export import SimpleSpanProcessor
from opentelemetry.sdk.resources import SERVICE_NAME, Resource

endpoint = "https://otlp.getbluejay.ai/v1/traces"
resource = Resource.create({SERVICE_NAME: "<your_service_name>"})

tracer_provider = trace_sdk.TracerProvider(resource=resource)
headers = {
    "X-API-KEY": "[BLUEJAY PROVISIONED API KEY]"
}

# utilize this tracer_provider with any OpenTelemetry/OpenInference instrumentation
tracer_provider.add_span_processor(SimpleSpanProcessor(OTLPSpanExporter(endpoint, headers=headers)))
OpenAI Example: 
from opentelemetry.instrumentation.openai import OpenAIInstrumentor

OpenAIInstrumentor().instrument(tracer_provider=tracer_provider)