Amazon Bedrock AgentCore Memory announces metadata for long-term memory

Amazon Bedrock AgentCore Memory Updates

Amazon Bedrock AgentCore Memory now supports metadata on long-term memory (LTM) records, enabling agents to tag, filter, and retrieve memories using structured attributes alongside semantic search.

You can define up to ten indexed keys per memory resource with support for STRING, NUMBER, and STRING_LIST types, and use different operator types to filter retrieval results.

Metadata can be attached to events at ingestion time or inferred automatically by the LLM based on extraction instructions you define on the memory resource. During ingestion, the LLM processes all events and determines how metadata is applied to the resulting memory records.

You define a metadata schema on the memory resource that includes indexed key definitions (key name, type, and optional allowed values) along with extraction instructions that guide the LLM on how to generate metadata from conversation content. With metadata filters on retrieval, agents can retrieve records by structured attributes like ticket number, priority, or date, eliminating irrelevant context and improving response accuracy.

What to do

• Define a metadata schema on your memory resource.
• Use extraction instructions to guide the LLM on generating metadata.
• Filter retrieval results using metadata to improve response accuracy.

Source: AWS release notes

If you need further guidance on AWS, our experts are available at AWS@westloop.io. You may also reach us by submitting the Contact Us form.