LangSmith can capture traces generated by CrewAI using OpenInference’s CrewAI instrumentation. This guide shows you how to automatically capture traces from your CrewAI multi-agent workflows and send them to LangSmith for monitoring and analysis.

Installation

Install the required packages using your preferred package manager:
pip install langsmith crewai openinference-instrumentation-crewai openinference-instrumentation-openai

Setup

1. Configure environment variables

Set your API keys and project name:
export LANGSMITH_API_KEY=<your_langsmith_api_key>
export LANGSMITH_PROJECT=<your_project_name>
export OPENAI_API_KEY=<your_openai_api_key>

2. Configure OpenTelemetry integration

In your CrewAI application, import and configure the LangSmith OpenTelemetry integration along with the CrewAI and OpenAI instrumentors:
from langsmith.integrations.otel import OtelSpanProcessor
from opentelemetry import trace
from opentelemetry.sdk.trace import TracerProvider
from openinference.instrumentation.crewai import CrewAIInstrumentor
from openinference.instrumentation.openai import OpenAIInstrumentor

# Get or create tracer provider
tracer_provider = trace.get_tracer_provider()
if not isinstance(tracer_provider, TracerProvider):
    tracer_provider = TracerProvider()
    trace.set_tracer_provider(tracer_provider)

# Add OtelSpanProcessor to the tracer provider
tracer_provider.add_span_processor(OtelSpanProcessor())

# Instrument CrewAI and OpenAI
CrewAIInstrumentor().instrument()
OpenAIInstrumentor().instrument()

3. Create and run your CrewAI application

Once configured, your CrewAI application will automatically send traces to LangSmith: This example includes a minimal app that defines agents and tasks, creates a crew, and runs it to produce traced activity.
from crewai import Agent, Task, Crew
from langsmith.integrations.otel import OtelSpanProcessor
from opentelemetry import trace
from opentelemetry.sdk.trace import TracerProvider
from openinference.instrumentation.crewai import CrewAIInstrumentor
from openinference.instrumentation.openai import OpenAIInstrumentor
import dotenv

# Load environment variables
dotenv.load_dotenv(".env.local")

# Configure OpenTelemetry
tracer_provider = trace.get_tracer_provider()
if not isinstance(tracer_provider, TracerProvider):
    tracer_provider = TracerProvider()
    trace.set_tracer_provider(tracer_provider)

# Add OtelSpanProcessor to the tracer provider
tracer_provider.add_span_processor(OtelSpanProcessor())

# Instrument CrewAI and OpenAI
CrewAIInstrumentor().instrument()
OpenAIInstrumentor().instrument()

# Define your agents
market_researcher = Agent(
    role="Senior Market Researcher",
    goal="Analyze market trends and consumer behavior in the tech industry",
    backstory="""You are an experienced market researcher with 10+ years of experience
    analyzing technology markets. You excel at identifying emerging trends and
    understanding consumer needs.""",
    verbose=True,
    allow_delegation=False,
)

content_strategist = Agent(
    role="Content Marketing Strategist",
    goal="Create compelling marketing content based on research insights",
    backstory="""You are a creative content strategist who transforms complex market
    research into engaging marketing materials. You understand how to communicate
    technical concepts to different audiences.""",
    verbose=True,
    allow_delegation=False,
)

data_analyst = Agent(
    role="Data Analyst",
    goal="Provide statistical analysis and data-driven insights",
    backstory="""You are a skilled data analyst who can interpret complex datasets
    and provide actionable insights. You excel at finding patterns and trends
    in data that others might miss.""",
    verbose=True,
    allow_delegation=False,
)

# Define your tasks
research_task = Task(
    description="""Conduct comprehensive research on the current state of AI adoption
    in small to medium businesses. Focus on:
    1. Current adoption rates and trends
    2. Main barriers to adoption
    3. Most popular AI tools and use cases
    4. ROI and business impact metrics

    Provide a detailed analysis with supporting data and statistics.""",
    agent=market_researcher,
    expected_output="A comprehensive market research report on AI adoption in SMBs with data, trends, and insights.",
)

analysis_task = Task(
    description="""Analyze the research findings and identify key statistical patterns.
    Create data visualizations and provide quantitative insights on:
    1. Adoption rate trends over time
    2. Industry-specific adoption patterns
    3. ROI correlation analysis
    4. Barrier impact assessment

    Present findings in a clear, data-driven format.""",
    agent=data_analyst,
    expected_output="Statistical analysis report with key metrics, trends, and data-driven insights.",
    context=[research_task],
)

content_task = Task(
    description="""Based on the research and analysis, create a compelling marketing
    strategy document that includes:
    1. Executive summary of key findings
    2. Target audience personas based on adoption patterns
    3. Key messaging framework addressing main barriers
    4. Content recommendations for different business segments
    5. Campaign strategy to drive AI adoption

    Make the content actionable and business-focused.""",
    agent=content_strategist,
    expected_output="Complete marketing strategy document with personas, messaging, and campaign recommendations.",
    context=[research_task, analysis_task],
)

# Create and run the crew
crew = Crew(
    agents=[market_researcher, data_analyst, content_strategist],
    tasks=[research_task, analysis_task, content_task],
    verbose=True,
    process="sequential"  # Tasks will be executed in order
)

def run_market_research_crew():
    """Run the market research crew and return results."""
    result = crew.kickoff()
    return result

if __name__ == "__main__":
    print("Running CrewAI market research process...")
    output = run_market_research_crew()
    print("\n" + "="*50)
    print("CrewAI Process Output:")
    print("="*50)
    print(output)

Advanced usage

Custom metadata and tags

You can add custom metadata to your traces by setting span attributes in your CrewAI application:
from opentelemetry import trace

# Get the current tracer
tracer = trace.get_tracer(__name__)

def run_market_research_crew():
    with tracer.start_as_current_span("crewai_market_research") as span:
        # Add custom metadata
        span.set_attribute("langsmith.metadata.crew_type", "market_research")
        span.set_attribute("langsmith.metadata.agent_count", "3")
        span.set_attribute("langsmith.metadata.task_complexity", "high")
        span.set_attribute("langsmith.span.tags", "crewai,market-research,multi-agent")

        # Run your crew
        result = crew.kickoff()
        return result

Combining with other instrumentors

You can combine CrewAI instrumentation with other instrumentors by adding them to your setup:
from openinference.instrumentation.crewai import CrewAIInstrumentor
from openinference.instrumentation.openai import OpenAIInstrumentor
from openinference.instrumentation.dspy import DSPyInstrumentor

# Initialize multiple instrumentors
CrewAIInstrumentor().instrument()
OpenAIInstrumentor().instrument()
DSPyInstrumentor().instrument()