AWS Bedrock provides access to high-performance foundation models through a unified API. Langtrace integration with AWS Bedrock allows you to capture and analyze interactions with various foundation models, helping you optimize and monitor your AI applications.

​
Available Models

Before getting started, you’ll need to request access to the specific foundation models you plan to use through the AWS Console. Visit Supported Models for a list of models available to use with the converse API.

​
Setup

​
Install Langtrace SDK

Install and initialize the SDK in your code.

​
Install AWS SDK

Install the AWS SDK for Python/Typescript:

pip install boto3

​
Usage Example

import boto3
from langtrace_python_sdk import langtrace

# Initialize Langtrace
langtrace = Langtrace(
    api_key="LANGTRACE_API_KEY",
)

# Initialize Bedrock client
client = boto3.client(
    service_name='bedrock-runtime',
    region_name='us-east-1',
    aws_access_key_id=<AWS_ACCESS_KEY_ID>,
    aws_secret_access_key=<AWS_SECRET_ACCESS_KEY>,
)

# Example usage
conversation = [
    {
        "role": "user",
        "content": [{"text": "Write a story about a magic backpack."}],
    }
]
response = client.converse(
    modelId=model_id,
    messages=conversation,
    inferenceConfig={"maxTokens":4096,"temperature":0},
    additionalModelRequestFields={"top_k":250}
)
response_text = response["output"]["message"]["content"][0]["text"]
print(response_text)

​
Troubleshooting

​
Common Issues

  1. Model Access: If you receive a “ResourceNotFoundException”, ensure you’ve requested and been granted access to the model.
  2. Trace Capture: If traces aren’t appearing, verify you’re using the converse API and the Langtrace SDK is properly initialized.
  3. Region Availability: Some models may not be available in all AWS regions. Check the AWS Bedrock documentation for region availability.

​
Support

For additional help: