Customers can now search video by meaning with zero infrastructure to build or manage
Joint customer Iconik makes semantic search a native capability of its platform and becomes TwelveLabs’ first Premier Tier ecosystem partner
SAN FRANCISCO, Sept. 10, 2026 (GLOBE NEWSWIRE) -- TwelveLabs, a leading video intelligence company, today announced that its Marengo model is now available as an embedding model inside Amazon Bedrock Managed Knowledge Base. For the first time, customers can search video the way they search text: by meaning. A user can type "show me the penalty kick in the second half" and get the exact moment from thousands of hours of footage, with nothing to build or manage.
Until now, making video searchable by meaning required months of engineering: building pipelines to extract frames, transcribe audio, generate embeddings, and stitch it all together with a custom database. Most teams never finish the project. With Marengo in Amazon Bedrock Knowledge Base, that entire process is replaced by a single managed service. Customers connect their video library, and search just works.
"Studios, sports leagues, broadcasters, and more are sitting on massive libraries of video that hold incredible value, but until now, unlocking that value meant huge amounts of custom engineering to build search and retrieval infrastructure," said Jose Kunnackal John, Director, Amazon Agentic AI. "With TwelveLabs' Marengo model now available natively in Amazon Bedrock Managed Knowledge Base, customers can simply point to their video archives and start searching by meaning, with everything from ingestion to indexing managed by AWS, saving time and making it simple to get the most out of their videos."
With Marengo in Amazon Bedrock Managed Knowledge Base, customers can:
- Search video, audio, and images using natural language. Ask a question in plain English and get the exact moment, image or clip.
- Get started in minutes, not months. Connect a video source, such as Amazon Simple Storage Service (Amazon S3), and Amazon Bedrock Managed Knowledge Base handles ingestion, indexing, and search automatically. No vector database to provision, no pipelines to build.
- Search across what's seen and what's said simultaneously. Marengo understands visual scenes, on-screen text, speech, and audio together, with ranking handled automatically.
- Combine meaning-based search with metadata filters. Narrow results by date, category, or tag alongside semantic queries.
- Keep all data inside their own AWS account. No video leaves the customer's environment, meeting the requirements of regulated industries like financial services, healthcare, and government.
“By collaborating closely with AWS, we’ve been able to eliminate the biggest barrier to deploying world-class video search capabilities,” said Danny Nicolopoulos, Head of Strategic Partnerships at TwelveLabs. “If your video is in Amazon S3, you can now use Marengo’s multimodal search capabilities immediately, without writing code or deploying vector search infrastructure. This offering represents a huge leap forward for semantic video search.”
Customers are already putting the integration to work. Iconik, the cloud-native, API-first media asset management platform from Backlight, is among the first using Marengo through Amazon Bedrock Managed Knowledge Base. The integration will give Iconik’s customers access to TwelveLabs’ multimodal video search capabilities on AWS, making TwelveLabs-powered semantic search a native capability of the Iconik platform.
Iconik became the first TwelveLabs Ecosystem Partner to achieve Premier Tier status, the program’s highest designation. This underscores Iconik’s commitment to video intelligence.
"Unlocking real value from every hour of footage — helping customers find, use and monetize their content faster — is something we're constantly investing in at Iconik,” said Ed Laczynski, Head of Strategic Partnerships at Backlight. “This partnership furthers that mission, giving our customers best-in-class video understanding from TwelveLabs the moment they need it. It's also proof of what we've always believed: customers get the most value when they can leverage any technology that fits how they work, and we're excited to keep building that choice for them."
TwelveLabs will showcase Marengo in Amazon Bedrock Managed Knowledge Base and its Iconik integration at IBC 2026 in Amsterdam, September 11–14. The company will conduct live demos in booth 5.D82.
To learn more about the integration, please visit https://www.twelvelabs.io/blog/marengo-amazon-bedrock-managed-knowledge-base.
About TwelveLabs
TwelveLabs is the world's most powerful video intelligence platform, that enables machines to perceive, understand, and reason about video the way humans do. Its architecture unifies perception, knowledge, and reasoning into a single system that compounds in value over time. From semantic search to automated analysis and multimodal understanding, TwelveLabs empowers developers, enterprises, and creatives to transform video from a storage cost into a strategic asset across industries including media, entertainment, advertising, government, security, and automotive. The company is headquartered in San Francisco with operations in Seoul, New York, Los Angeles, and London. For more information, visit www.twelvelabs.io.
About Iconik by Backlight
Iconik is the creative operations platform that replaces fragmented creative toolkits with a single AI-powered system for finding, reviewing, and publishing media assets at scale. More than a modern media asset management (MAM) system, Iconik integrates with cloud-based and on-prem storage, automates tasks like metadata generation, and connects with existing workflows. It adds a layer of media intelligence to all assets, giving teams the flexibility they need to grow their media business without losing creative control. For more information, visit https://www.iconik.io.
