AWS introduces AgentCore to simplify AI agent deployment

AWS has unveiled a preview of Amazon Bedrock AgentCore, a suite of enterprise-grade services designed to simplify the deployment and operation of AI agents at scale. Announced on July 16, the initiative aims to reduce the time developers spend building foundational infrastructure for identity controls, session management, and observability.

Key Features of AgentCore

AgentCore services can be hosted on Amazon Bedrock or other platforms, offering developers flexibility in deployment. The services include:

• AgentCore Runtime: A low-latency serverless environment supporting any agent framework, including open-source tools and models.
• AgentCore Memory: Manages session and long-term memory to provide context for AI agents.
• AgentCore Observability: Offers step-by-step visualization of agent execution for debugging and troubleshooting.
• AgentCore Identity: Securely grants agents access to AWS and third-party services like GitHub and Slack.
• AgentCore Gateway: Transforms APIs and AWS Lambda functions into agent-ready tools.
• AgentCore Browser: Scales web automation workflows with managed browser instances.
• AgentCore Code Interpreter: Provides an isolated environment for running agent-generated code.

Compatibility and Pricing

AgentCore works with frameworks such as CrewAI, LangGraph, and LlamaIndex, as well as any foundation model inside or outside Amazon Bedrock. Developers can use the services for free until September 16, 2025, after which standard AWS pricing will apply. Usage billing for AgentCore begins on September 17, 2025.

Use Cases

AgentCore is ideal for applications like:

• Customer support
• Workflow automation
• Innovative AI-powered experiences

By offering these tools, AWS aims to accelerate the adoption of AI agents in enterprise environments, reducing the overhead of building custom infrastructure.