How Condé Nast built multimodal video discovery with Amazon Bedrock
aws ·
Condé Nast partnered with the AWS Generative AI Innovation Center to develop an AI-powered multimodal video discovery solution, addressing significant operational inefficiencies in their editorial workflow. Previously, teams spent an average of 250 minutes manually searching through a library of over 140,000 videos using only titles and descriptions. The new system utilizes Amazon Bedrock and the TwelveLabs Marengo embedding model to perform semantic searches across visual, audio, and transcript data simultaneously. By decoupling the ingestion pipeline from the query serving tier, the architecture ensures high availability and scalability across multiple Availability Zones. This intent-based search allows editors to find specific clips with precise timestamps using natural language or image queries, significantly reducing manual review efforts. As a result, Condé Nast achieved a 99.2 percent reduction in content discovery time, lowering it to under two minutes per task. The initiative also generated approximately $800,000 in estimated annual operational savings and accelerated revenue capture by speeding up responses to advertiser requests. The solution demonstrates how large media organizations can leverage multimodal AI to unlock value from extensive video archives.
aws