Amazon Web Services said Amazon Bedrock AgentCore is generally available and said the AgentCore software development kit has seen more than 1 million downloads from customers.
AgentCore is the latest entry into the agentic AI platform race. Google Cloud launched Gemini Enterprise, which absorbs what used to be Agentspace. Salesforce outlined Agentforce 360, the fourth version of Agentforce in 12 months. Those agentic AI platforms landed in recent days, but there are multiple vendors such as ServiceNow, Boomi and UiPath all looking to be the management layer for AI agents.
What AWS is hoping to do with AgentCore, initially outlined at AWS Summit New York, is provide a horizontal stack that can enable customers to deploy, operate and secure AI agents with all the necessary building blocks such as tools, memory, data and workflows. AWS considers AgentCore to be a key platform for operationalizing AI agents.
- AWS launches Amazon Quick Suite, aims to automate business workflows
- The AWS AI Strategy: Playing the Long Game Infrastructure-Style
- AWS' AI strategy: Jassy's long talking and the big picture
- AWS Q2 sales growth 17.5%, nears $124 billion annual revenue run rate
- AWS launches Bedrock Agent Core, custom Nova models
- AWS' practical AI agent pitch to CxOs: Fundamentals, ROI matter
By operating horizontally, AWS sees AI agents as a natural extension to its Bedrock platform, which provides model choices.
In a recent briefing, AWS noted that AgentCore is designed to give enterprises the flexibility to start with any model or agent framework. Early customers include Cox Automotive, Druva, Cohere Health, Ericsson, Sony and Thomson Reuters to name a few. AWS has also surrounded AgentCore with big partners such as Accenture, Deloitte and Salesforce.
Key points about AgentCore include:
- AgentCore includes composable services that can be used together or independently.
- The platform supports multiple AI agent frameworks such as CrewAI, Google ADK, LangGraph, LlamaIndex, OpenAI Agents SDK and Strands Agents and models on Amazon Bedrok, OpenAI or Gemini.
- AgentCore includes AgentCore Code Interpreter, AgentCore Browser and AgentCore Gateway, which takes existing APIs and AWS Lambda functions and makes them agent friendly tools with connections to third-party data and Model Context Protocol servers.
- AWS has also added AgentCore Identity, which provides authentication and authorization for AI agents.
- Dashboards and tracking is provided via AgentCore Observability, which integrates with various monitoring tools.
- AgentCore Runtime automatically scales agent workloads and security so each agent session has its own isolated computing environment. AgentCore supports virtual private cloud deployments.
According to AWS, the goal of AgentCore is to get from pilot to production faster in a way that is secure with little friction for developers.
