Amazon Bedrock Now Supports Semantic Video Search

Amazon Bedrock Knowledge Bases now supports TwelveLabs Marengo Embed 3.0, enabling natural language search across video, audio, and image content. The fully managed service eliminates complex pipelines by handling segmentation, transcription, embedding, and retrieval automatically. Marengo Embed 3.0 encodes video, audio, images, and text into a unified 512-dimensional vector space. Users can query footage with plain language, such as finding penalty kicks in a soccer match, with results returning ranked segments including timestamps and source metadata.