This integration demonstrates how to use Strands Agents with OpenInference and Arize AI for observability. The example builds a restaurant assistant agent that helps customers with restaurant information and reservations.
- Python 3.10+
- AWS account with appropriate permissions
- Anthropic Claude 3.7 enabled on Amazon Bedrock
- IAM role with permissions to create Amazon Bedrock Knowledge Base, Amazon S3 bucket and Amazon DynamoDB
- Arize AI account with API key and Space ID
-
Clone the repository:
git clone https://github.com/strands-agents/samples.git cd 03-integrations/Openinference-Arize -
Run the notebook:
Arize-Observability-openinference-strands.ipynb -
Update the Arize AI credentials in the notebook:
- Get your API key and Space ID from Arize AI dashboard
- Replace the placeholder values in the notebook
-
View the traces in the Arize AI dashboard
After running the agent, you can explore the traces and set up monitoring in Arize AI:
-
Project Traces
-
Filtering Traces
-
Filtering by Model ID
-
Trace Details
-
Execution Graph
-
Inspecting Execution Paths
-
Session Information
-
Pre-built Monitors
-
Monitor View
For more information on monitoring, see the Arize documentation on monitoring.
When you're done experimenting, please clean up the AWS resources:
sh cleanup.sh
- How to set up OpenInference with Arize AI for Strands Agents
- How to use the custom
StrandsToOpenInferenceProcessorto convert telemetry - How to build a functional restaurant assistant with Strands
- How to visualize and analyze agent behavior in Arize AI
- How to set up monitoring for your agent in production