Verified, then logged.
Sun 22 Nov 2026 · 00:20 KST / 15:20 UTC한국어 | ENSearch /

Breaking02 / IT

AWS links user memory to Bedrock search

The integration lets the service restore and save conversations while keeping document access controls separate from user memory identifiers.

Amazon Web Services (AWS) introduced user memory integration for search in Amazon Bedrock, its AI application platform. It presented traces from a sample application run but did not disclose usage figures.

AWS said the integration connects AgentCore memory to search, allowing information extracted from earlier conversations to shape queries sent to the knowledge base, a collection of documents used to ground answers.

According to AWS, the integration is optional and remains compatible with existing requests. Applications enable it by adding memory configuration to a search request.

The AWS example separates document permissions from memory identifiers. It uses a verified email address to determine document access and a stable, unique user identifier to keep each user's memory separate.

AWS designed the integration to replace the conversation-history storage step in its earlier example. The service restores the conversation before processing a request and saves the exchange after generating an answer.

AWS did not publish response latency figures for the example run.