Building agentic workflows with SageMaker AI and Bedrock AgentCore
Learn how to combine OpenAI-compatible endpoints on Amazon SageMaker AI with Amazon Bedrock AgentCore runtime to build a multi-agent workflow where each specialized agent uses the model best suited to its job. This post also shows how to get token-level observability from SageMaker endpoints that Strands Agents does not instrument by default.
This post demonstrates how to build agentic workflows using Amazon SageMaker AI and Amazon Bedrock AgentCore. The architecture combines managed foundation models from SageMaker AI with domain-specific models from Bedrock, enabling cost optimization, data residency, and model flexibility in a single production-ready system. Qwen 3.5 9B runs on SageMaker AI as the financial analysis agent, while Claude Haiku 4.5 and Claude Sonnet 4.6 run on Bedrock as the orchestrator and budget agents, respectively.
By integrating these models, users can classify user intent, handle budget breakdowns, and perform stock analysis and portfolio construction with specialized agents. To get token-level observability from SageMaker endpoints, which Strands doesn't provide, a custom httpx.Auth subclass is used for refreshing bearer tokens in the OpenAI-compatible API.
This setup allows for a streamlined, end-to-end workflow with the desired combination of models and observability.
Written by urgent.news from AWS Machine Learning's reporting — not their text. Machine-written — it may contain errors, so check the original before relying on it.
