New components and capabilities, shipped every few weeks. Filter by category or browse the full history.
AI, quality monitoring, standards conversion, and next-generation delivery protocols.
Las Vegas · April 2026
"Norsk Reasoning represents a fundamental shift in how live production operates. AI that can observe a live stream, understand what's happening across both video and audio, and take action in real time eliminates entire categories of manual monitoring work." Adrian Roe, CEO — Norsk
A multi-modal AI system natively integrated into the Norsk pipeline that moves beyond passive analysis to autonomous action. Operators describe detection objectives in natural language; a planning model structures those into a detection specification that a faster evaluation model executes against the live stream in real time. The system processes video and audio simultaneously, maintains temporal context across the stream, and can autonomously switch camera sources, trigger recordings, fire webhooks, and update graphics — with human override always available.
Supported models include GPT-4o, Gemini, Claude, and local alternatives through a unified, vendor-neutral interface.
Native MoQ support delivers sub-second latency at CDN scale — matching WebRTC's responsiveness while supporting relay-based delivery to large audiences. Demonstrated in a joint demo with Cloudflare, showing live media delivered over Cloudflare's global MoQ relay network from a Norsk Studio workflow.
Support for Common Media Server Data – Media Quality Assessment, an emerging SVTA standard that embeds objective video quality scores into HTTP response headers as content moves through the delivery chain. Preserves quality scores from encoder to CDN to player, enabling quality-based switching decisions across the entire workflow — not just at a single monitoring point.
The MediaStore component brings Time-Addressable Media Store natively into Norsk Studio workflows. Live content becomes immediately queryable as it's captured, eliminating the traditional boundary between live production and VOD. TAMS — developed by BBC R&D and adopted by AWS, Adobe, Sky, and others — stores segmented media in cloud object storage with HTTP API access indexed by timeline position.
Support for the Media eXchange Layer, a Linux Foundation-hosted open standard for shared-memory media exchange between software applications within the same server or cluster. MXL 1.0.0 enables sub-millisecond media transfer between components — a 20× latency improvement over intra-server ST 2110 workflows — enabling tighter integration across the software-defined broadcast stack.
An early look at Norsk Playout — a scheduled playout system designed to switch seamlessly between live feeds, VoD assets, and TAMS-stored media. Human operators and the Norsk Reasoning AI layer provide parallel oversight and control, combining automation with human-in-the-loop reliability. General availability to follow later in 2026.