Observing agents built with CrewAI

Make sure that the following prerequisites are met:

• Python 3.8 or later

• An LLM API key (Anthropic, OpenAI, or any LLM provider supported by CrewAI)

• Instana configured for your application, see Getting started

To instrument your CrewAI application with Instana, complete the following steps:

1. Install the required packages for CrewAI and OpenLLMetry.

   pip install crewai crewai-tools traceloop-sdk

2. Set your LLM provider credentials and disable CrewAI's internal telemetry.

   export ANTHROPIC_API_KEY="<your-api-key>"
   export CREWAI_DISABLE_TELEMETRY=true

   Note:

   Replace ANTHROPIC_API_KEY with your LLM provider's environment variable if using a different provider (e.g., OPENAI_API_KEY for OpenAI).

   Disabling CrewAI's internal telemetry ensures all traces are captured through OpenLLMetry for consistent observability.

3. Create your CrewAI application.

   In the following example, two agents are created (a researcher and a writer) to collaborate for a task:

   from crewai import Agent, Task, Crew, Process
   from crewai_tools import SerperDevTool
   from traceloop.sdk import Traceloop
   from traceloop.sdk.decorators import workflow
   
   # Initialize OpenLLMetry
   Traceloop.init(app_name="crewai_demo")
   
   # Initialize search tool
   search_tool = SerperDevTool()
   
   # Define agents
   researcher = Agent(
       role="Senior Research Analyst",
       goal="Uncover cutting-edge developments in AI and data science",
       backstory="You are a Senior Research Analyst at a leading tech think tank.",
       verbose=True,
       allow_delegation=False,
       tools=[search_tool]
   )
   
   writer = Agent(
       role="Tech Content Strategist",
       goal="Craft compelling content on tech advancements",
       backstory="You are a renowned Tech Content Strategist, known for insightful and engaging articles on technology and innovation.",
       verbose=True,
       allow_delegation=False,
       tools=[search_tool]
   )
   
   # Define tasks
   task1 = Task(
       description="Perform an in-depth analysis of the following topic: {topic}",
       expected_output="Comprehensive analysis report in bullet points",
       agent=researcher
   )
   
   task2 = Task(
       description="Using the insights from the researcher's report, develop an engaging blog post that highlights the most significant advancements",
       expected_output="A compelling 3-paragraph blog post formatted as markdown",
       agent=writer
   )
   
   # Create the crew
   crew = Crew(
       agents=[researcher, writer],
       tasks=[task1, task2],
       verbose=True,
       process=Process.sequential
   )
   
   # Wrap crew execution in a workflow for better tracing
   @workflow(name="crewai_workflow")
   def run_crew(topic):
       return crew.kickoff(inputs={"topic": topic})
   
   # Execute the crew
   if __name__ == "__main__":
       topics = ["Artificial Intelligence", "Machine Learning", "Neural Networks"]
   
       for topic in topics:
           result = run_crew(topic)
           print(f"\nCompleted analysis for: {topic}\n")

4. Run your CrewAI application.

   python crewai_app.py

   CrewAI generates output showing the agents' collaboration:

   

5. View data in Instana

   After running your application, data will appear in the Instana Gen AI observability dashboard:

   

Troubleshooting information on agents built with CrewAI.

• Explore other AI agent frameworks supported by Instana

• Learn about cost calculation for your LLM usage

• Set up alerts for your CrewAI workflows