Eluvio Will Demonstrate AI Video Workflows That Avoid File Copies at IBC

Eluvio will demonstrate its Video Intelligence platform at IBC2026 in Amsterdam on September 11–14, putting its newest pitch to broadcasters and rights holders on display: AI search, automated clips and agent-led workflows that operate on source video without creating separate media copies. The immediate stakes are operational. Eluvio says analysis results can move straight into streaming, publishing and rights-management workflows rather than beginning another transfer and transcoding cycle.

The platform processes AI inline, meaning it analyzes frames directly on source files held in Eluvio’s Content Fabric. It can run Eluvio-developed, open-source and third-party models simultaneously on the same material. Eluvio says this design produces frame-accurate labels, embeddings, transcripts and other metadata without exporting media into a separate AI pipeline.

From video analysis to a usable media output

The IBC features fall into three connected layers. Natural-language AI Search and Q&A is meant to query a video library in ordinary language. AI Moment Detection and Clip Generator is intended to find moments and create highlights from live or archived content. Agentic Workflow Orchestration links AI agents, human review steps and downstream systems in a configurable pipeline.

• Metadata tagging, cataloging, scene detection, content moderation, highlights and clips are among the supported workflows.
• The platform includes 17 built-in models and processors spanning analysis, segmentation, summarization, transcription, identification, search and generation, according to Eluvio.
• Its runtime is designed to accept open-source, customer-specific, third-party and Eluvio-developed models through an extensible API.

Why the zero-copy design is central to the pitch

Eluvio says AI-derived metadata and clips can be used directly by its streaming, publishing and rights-management infrastructure, with no additional file transfer or re-transcoding. Its stated alternative is a workflow in which a media file is copied out for analysis and then prepared again for use elsewhere. The company says its approach instead lets metadata and video references drive personalized outputs and distribution from within the Fabric.

That distinction is especially relevant to Eluvio’s live-sports demonstration. The company plans to show dynamic 9:16 vertical video generated from a 16:9 source using motion analysis, focus tracking and per-player jersey identification. It says the system determines a focal point by shot, then produces vertical presentations without making a second encoded stream or pre-generating separate vertical assets.

An MCP connection turns chat prompts into workflow requests

Eluvio is also introducing a Model Context Protocol, or MCP, API. It is intended to let ChatGPT, Claude, third-party applications and other agentic systems invoke Content Fabric and Eluvio AI functions through natural-language interfaces. Eluvio says one workflow can run models, update vector indexes, search video, create transcripts, identify key moments, build chapters and compose clips.

The agent connection does not remove the need for review or permissions. Eluvio’s announced orchestration layer explicitly includes human review steps, while the company says MCP workflows operate through authorized functions under the Content Fabric’s tenant and owner-controlled permissions. Those details describe the intended controls; IBC demonstrations will be the nearer test of how the pieces work together across live video and library workflows.