Amazon Bedrock AgentCore Memory now supports direct ingestion to long-term memory

Amazon Bedrock AgentCore Memory now offers the IngestData API, enabling developers to submit content directly for long-term memory extraction without first storing it as a short-term memory event. The API processes both conversational payloads and JSON-based events, routing them through configured long-term memory strategies. Processed memory records become immediately retrievable using standard retrieval operations, eliminating the previous dependency on short-term memory storage. This change streamlines memory management workflows for developers working with long-term retention needs.

Previously, all content required conversion into a short-term memory event before extraction into long-term memory records could occur. The new IngestData API removes this mandatory intermediate step, allowing long-term memory strategies to operate independently of short-term memory workflows. Developers can now ingest behavioral events, activity logs, and system events directly into long-term memory, expanding use cases for persistent memory retention. The API maintains compatibility with existing extraction pipelines while reducing processing latency.

The IngestData API supports optional metadata fields that integrate with the same extraction pipeline used by the CreateEvent operation. After processing, developers can validate extraction results using ListMemoryRecords or RetrieveMemoryRecords, enabling verification of stored memory content. Real-time monitoring is available through Kinesis for streaming notifications, while failed extractions can be reprocessed via ListMemoryExtractionJobs. This provides operational flexibility for managing memory ingestion workflows.

To implement the new functionality, developers should consult the Amazon Bedrock AgentCore Developer Guide for Direct ingestion to long-term memory instructions. The IngestData API is currently available in all AWS Regions where Amazon Bedrock AgentCore Memory is supported, ensuring broad accessibility for existing and new deployments.