ScyllaDB Adds Native Vector Search to DynamoDB-Compatible API, Targeting AI Workloads and Cost Savings

New updates have been reported about ScyllaDB.

ScyllaDB has introduced native vector search for its DynamoDB-compatible Alternator API, positioning itself as an integrated alternative to the common DynamoDB plus OpenSearch stack for semantic and similarity search. By embedding vector capabilities directly in the database, ScyllaDB aims to remove the operational and cost burden of exporting data to S3, synchronizing via DynamoDB Streams, and maintaining a separate search cluster and API for AI-driven queries.

CEO Dor Laor argues that ScyllaDB’s approach cuts complexity, latency, and infrastructure expense compared with Amazon’s recommended “zero ETL” pattern, which still requires coordination between two systems and can suffer from sync failures, data drift, and reindexing overhead. The new vector search runs on ScyllaDB Cloud, requires no code changes for existing DynamoDB applications, and is marketed as delivering materially lower total cost of ownership for real-time AI workloads such as semantic search, retrieval-augmented generation pipelines, and recommendation engines.