Strands-Powered AWS Update Brings Three-Call Agent Deployment
Open Source For You, April 28,2026
AWS launches AgentCore harness based on Strands framework, enabling autonomous agent deployment in three API calls.
Amazon Web Services has expanded its open-source agent strategy with a new AgentCore harness built on the Strands framework, allowing developers to deploy autonomous agents through configuration with just three API calls rather than complex orchestration code.
The model-agnostic architecture handles reasoning, tool selection, action execution and response streaming, supporting Amazon Bedrock models alongside OpenAI and Google Gemini while maintaining compatibility with frameworks like LangGraph, LlamaIndex and CrewAI.
AWS is positioning infrastructure abstraction as the key differentiator for enterprise agentic AI adoption, offering the AgentCore CLI, persistent filesystem for long-running agents, and coding agent skills at no additional charge. Available in preview across four regions, this release strengthens AWS's competitive position against similar offerings from Google, Microsoft and OpenAI, though some limitations remain around regional restrictions and configuration complexity.