AWS Combines SageMaker AI with Bedrock AgentCore for AI Workflows

TL;DR. AWS details an architecture combining Amazon SageMaker AI and Amazon Bedrock AgentCore to build multi-agent LLM workflows. - This approach allows specialized agents to use different models, optimizing for cost and domain specificity. - The workflow integrates models like Qwen 3.5 9B with Claude 4.5/4.6, routing tasks based on user intent. - It also provides token-level observability from SageMaker endpoints, which Strands Agents lacks by default.

• AWS presented a method for integrating Amazon SageMaker AI with Amazon Bedrock AgentCore.
• This integration enables the creation of multi-agent workflows where each agent can leverage different LLMs.
• The architecture supports cost optimization, data residency, and model flexibility in production environments.
• It demonstrates using Qwen 3.5 9B on SageMaker AI alongside Claude models on Bedrock.
• The setup provides advanced observability for SageMaker endpoints within the agent framework.

Back to QLANKR News