The Delta Protocol accelerator, led by DAZN with broadcast and technology partners, has built a working end-to-end proof of concept for AI-native live delivery. The system analyses a sports feed and produces compact descriptions of players, weather, crowd response and meaningful scene changes.
Those delta instructions are synchronised to the video frame and delivered to the client beside the conventional feed. A client SDK can then alter overlays, personalised moments or picture enhancement without requiring the service to encode and distribute a separate complete video variant for every audience segment.
Google and NeoViews contribute scene understanding, Eluvio keeps the intelligence track aligned with video, and V-Nova is being evaluated as an alternative analysis feed. The IBC demonstration uses football and shows two devices producing different experiences from the same live event.
The project’s longer-term ambition to make the video path optional remains speculative. Real-time model errors, latency, client compute, rights enforcement and security need far more evidence. The demonstrated result is a synchronised semantic sidecar and client overlay, not full neural reconstruction of a live broadcast.
