Running agents in Amazon Bedrock AgentCore

"Amazon Bedrock AgentCore provides an enterprise-grade infrastructure and operations layer for deploying and managing AI agents at scale. AgentCore itself is completely agnostic about models, frameworks, and integrations, although its starter kit CLI only supports the most prominent of these."

"That CLI can generate agents using Amazon Bedrock, Anthropic, Google Gemini, and OpenAI models, using Strands, LangGraph, Microsoft Autogen, OpenAI Agents SDK, Google Agent Development Kit, and CrewAI frameworks, and using Amazon AgentCore Memory, AgentCore Observability, and AgentCore Gateway integrations."

"AgentCore's core services include a runtime, memory (both short-term and long-term), a gateway, identity management, a sandboxed code interpreter, a cloud-based browser, observability, an evaluation service, and a policy capability that runs outside the agent."

Amazon Bedrock AgentCore is an enterprise infrastructure and operations layer designed for deploying and managing AI agents at scale. The platform is model, framework, and integration agnostic, though its starter kit CLI supports prominent options including Amazon Bedrock, Anthropic, Google Gemini, and OpenAI models. It works with frameworks like Strands, LangGraph, Microsoft Autogen, OpenAI Agents SDK, Google Agent Development Kit, and CrewAI. Core services include runtime, short-term and long-term memory, gateway, identity management, sandboxed code interpreter, cloud-based browser, observability, evaluation service, and external policy capabilities. Direct competitors include Google Cloud Agent SDK, Azure AI Foundry Agents, and Databricks Agent Bricks, while additional competitors and collaborators include OpenAI Agents SDK, LangChain/LangGraph, CrewAI, and SmythOS.

Read at InfoWorld