Amazon Bedrock AgentCore Memory Enhances AI Agent Recall
Amazon Web Services has launched AgentCore Memory, a new service for Amazon Bedrock designed to help AI agents remember conversation history and user preferences.

Amazon Web Services (AWS) has introduced AgentCore Memory, a new component within its Amazon Bedrock platform. This service addresses a common challenge in artificial intelligence applications: the inability of large language models to retain context across interactions.
AgentCore Memory allows developers to easily build context-aware AI agents by managing the underlying memory infrastructure. This eliminates the need for developers to repeatedly construct custom memory systems for each application, a process that was previously complex and time-consuming.
The service provides both short-term working memory for immediate conversational context within a session and long-term intelligent memory for persistent insights and preferences across multiple sessions. This allows AI agents to maintain context, learn from user interactions, and deliver more personalized experiences.
Unveiled at AWS Summit New York City, AgentCore Memory is designed to streamline the development of more sophisticated AI agents. It integrates with existing agent-building tools, enabling agents to build upon past interactions and reducing the need to repeatedly ask users for the same information.