Keep your existing observability tool

Add Judgment as a second destination alongside Logfire, LangSmith, Braintrust or Langfuse.

All four tools build on OpenTelemetry, so Judgment can be one more span processor on the same tracer provider, and both tools get the same spans. This doesn’t import traces you already stored in another tool; if you need history moved over, contact support.

What Judgment reads from each tool

Tool	How you add Judgment	What Judgment’s ingest maps from the tool’s own attributes
Logfire	`additional_span_processors` in `logfire.configure()`	`logfire.msg` becomes the span name
LangSmith	LangSmith OpenTelemetry mode on a shared provider	`langsmith.span.kind` becomes the span kind
Braintrust	`BraintrustSpanProcessor` next to Judgment’s on a shared provider	model, provider, prompt and completion tokens, cost
Langfuse	A shared provider passed to `Langfuse(tracer_provider=...)`	Nothing. Judgment has no Langfuse mapper, so Langfuse is documented here as a destination only

Standard OpenTelemetry GenAI attributes (gen_ai.*) and OpenInference attributes map the same way no matter which tool created the span. If a field you need isn’t mapped, add a mapper: see Attribute keys and mappers.

Build the Judgment span processor

Each example below adds this processor. It’s a standard OTLP exporter, so it needs the project ID rather than the name; see Send OTLP directly for where to find it.

pip install opentelemetry-sdk opentelemetry-exporter-otlp-proto-http

import os

from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk.trace.export import BatchSpanProcessor


def create_judgment_processor() -> BatchSpanProcessor:
    return BatchSpanProcessor(
        OTLPSpanExporter(
            endpoint="https://api.judgmentlabs.ai/otel/v1/traces",
            headers={
                "Authorization": f"Bearer {os.environ['JUDGMENT_API_KEY']}",
                "X-Organization-Id": os.environ["JUDGMENT_ORG_ID"],
                "X-Project-Id": os.environ["JUDGMENT_PROJECT_ID"],
            },
        )
    )

Both tools now receive the same data, so apply the same scrubbing and sampling policy you already use.

Logfire

import logfire
from judgment_otlp import create_judgment_processor

logfire.configure(
    service_name="lunch-order-agent",
    additional_span_processors=[create_judgment_processor()],
)
logfire.instrument_openai()

If logfire.configure() rejects additional_span_processors, upgrade logfire.

LangSmith

LangSmith’s OpenTelemetry mode uses the global provider when one is set. Set one that includes Judgment, and leave LANGSMITH_OTEL_ONLY unset so runs still go to LangSmith too. This needs langsmith[otel].

import os

from judgment_otlp import create_judgment_processor
from opentelemetry import trace
from opentelemetry.sdk.trace import TracerProvider

os.environ["LANGSMITH_OTEL_ENABLED"] = "true"
os.environ["LANGSMITH_TRACING"] = "true"

provider = TracerProvider()
provider.add_span_processor(create_judgment_processor())
trace.set_tracer_provider(provider)

Run this before your first chain or graph runs.

Braintrust

Install braintrust[otel].

from braintrust.otel import BraintrustSpanProcessor
from judgment_otlp import create_judgment_processor
from opentelemetry import trace
from opentelemetry.sdk.trace import TracerProvider

provider = TracerProvider()
provider.add_span_processor(BraintrustSpanProcessor(parent="project_name:lunch-order-agent"))
provider.add_span_processor(create_judgment_processor())
trace.set_tracer_provider(provider)

Langfuse

Langfuse’s Python SDK v3 and later accepts a tracer provider. Add Judgment to it and keep your Langfuse instrumentation as it is.

import os

from judgment_otlp import create_judgment_processor
from langfuse import Langfuse
from opentelemetry import trace
from opentelemetry.sdk.trace import TracerProvider

provider = TracerProvider()
provider.add_span_processor(create_judgment_processor())
trace.set_tracer_provider(provider)

langfuse = Langfuse(
    public_key=os.environ["LANGFUSE_PUBLIC_KEY"],
    secret_key=os.environ["LANGFUSE_SECRET_KEY"],
    tracer_provider=provider,
)