Eluvio has unveiled what it describes as a new architecture for “universal and dynamic” video intelligence, alongside the next generation of its Eluvio Video Intelligence Editor (EVIE), ahead of NAB Show 2026 in Las Vegas.
The Berkeley-based company said the new Eluvio AI platform is the first commercially available system to run multimodal AI analysis inline within the streaming media generation and distribution pipeline, allowing frame-accurate analysis of live and on-demand content without file copies, file movement or re-transcoding.
According to Eluvio, the architecture is designed to support live sports, events, studio archives and other premium video workflows by generating aligned labels, embeddings and metadata across video, audio, text, images and telemetry data.
The company said the system enables real-time inline inference on live streams as well as VOD assets, while also supporting derivative content creation including highlights, trailers, shorts, captions and synopses as zero-copy outputs.
A key part of the launch is the latest version of EVIE, which Eluvio is positioning as an AI-native, frame-accurate editing and workflow platform for multimodal content search, automatic highlights creation, short-form video generation and metadata production.
Eluvio is also previewing new EVIE Titles and Events beta features, bringing together search, summarisation and composition tools in a single user experience and API layer organised around either title libraries or live events. New functions include vertical video generation from 16:9 sources, live AI motion analysis and orchestration APIs intended to support natural language prompts, third-party chatbots and other agentic interfaces.
Among the features highlighted by Eluvio are deep content search across semantic, image, text and music inputs, automatic highlight generation using play-by-play and tracking data, and tools for producing social-ready clips and promotional assets.
The company said its platform remains open to third-party, open-source and custom AI models, with more than 15 built-in models and processors already available. Newly added models include focus and speaker detection, custom motion identification, OpenCLIP, ImageBind and an enhanced MediaPipe implementation.
Michelle Munson, CEO and co-founder of Eluvio, said media companies needed “operational reality and the right architecture” from AI, combining frame accuracy, live inference and zero-copy efficiency.
She added that EVIE and Eluvio’s orchestration APIs were designed to give media professionals a single pipeline from live streaming to progressive VOD, with in-place AI assistance for personalised highlights, topic-based channels and new derivative content workflows.
Eluvio said it will demonstrate the new capabilities at NAB Show 2026 from 19 to 22 April at the Las Vegas Convention Center, with exhibits at AI Innovation Pavilion Booth W2300E and West Hall Meeting Room W237.
In a separate announcement, the company also said it is making its Content Fabric “Bucharest Release” commercially available, positioning it as a unified platform for zero-copy broadcast distribution, D2C OTT streaming and inline intelligence.
