Build memory-aware AI agents with Agentic Retrieve on Amazon Bedrock Managed Knowledge Base
Machine Learning Blog
This article demonstrates how to build memory-aware AI agents using Amazon Bedrock Managed Knowledge Base with AgentCore memory integration, enabling personalized, context-aware responses while maintaining multi-user isolation and document access control.
- AgentCore memory stores conversation history and extracts long-term facts to personalize answers across sessions
- Document ACLs filter retrieval per user email, ensuring each user sees only authorized documents
- Memory namespaces keyed by user ID (sub) provide per-user isolation even with a shared memory resource
- IAM condition keys enforce access control as a second line of defense against misconfigured application code
- Cognito ID tokens supply verified email for document filtering and stable user ID for memory keying
- Session restore and persistence happen automatically within AgenticRetrieveStream, eliminating custom conversation storage
- Trace events reveal when the planner uses memory and whether retrieval succeeded or failed
The integration simplifies multi-tenant AI assistants by combining authentication, document filtering, conversation management, and memory extraction into a single service call.
The AWS News Feed is currently looking for gold sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.
Related articles
2026
2026
2026
2026
The AWS News Feed is currently looking for silver sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.