Build RAG-ready indexes from Amazon S3 with Amazon OpenSearch intelligent ingestion

This article announces Amazon OpenSearch intelligent ingestion, a fully managed serverless capability that automatically ingests and processes documents from Amazon S3 into RAG-ready indexes.

• Automatically parses PDFs, Word, and Markdown files while preserving layout, reading order, and tables
• Generates text descriptions of images using vision-language models to make visual content searchable
• Processes structured data (JSON, NDJSON, CSV) with optional semantic enrichment without pipeline code
• Supports Ingestion Jobs for one-time loads and Ingestion Streams for continuous synchronization
• Automatically chunks content, generates vector embeddings, and scales capacity based on workload
• Charges only for active processing with no idle capacity costs

Intelligent ingestion eliminates weeks of pipeline engineering by handling document discovery, parsing, image understanding, chunking, embedding, and indexing automatically within OpenSearch Serverless.