Refraction DAO and Livepeer Join Forces for Brussels Event

The gaming industry is shifting from pre-scripted content to real-time AI, fundamentally changing how 3 billion gamers worldwide experience games. **Key Changes:** - **Reactive NPCs**: Non-player characters that genuinely respond to player actions rather than following predetermined scripts - **Dynamic Worlds**: Game environments that adapt and change based on player behavior - **Personalized Experiences**: Each playthrough becomes unique and tailored to individual players This evolution moves beyond traditional AI agents that simply remember moves and adjust strategies. Real-time AI creates truly bespoke gaming experiences where every interaction matters and shapes the game world.

Livepeer orchestrators are experiencing record network fees as demand for decentralized video infrastructure grows. **Key Drivers:** - AI video processing and real-time streaming applications are increasing network usage - Projects like [DaydreamLiveAI](https://www.livepeer.org/) and Agent SPE integrations are creating high-frequency, low-latency video workloads - GPU compute demands for interactive pipelines are expanding beyond traditional transcoding **Impact on Orchestrators:** - Higher fee earnings from increased network utilization - More diverse and persistent workloads being routed to nodes - Better revenue opportunities for node operators **Network Health Indicators:** - Real adoption of the infrastructure - Sustainable demand from AI-powered applications - Self-reinforcing growth cycle The rise in fees signals genuine ecosystem growth, with AI agents, autonomous VTubers, and generative video workflows driving traffic to Livepeer's decentralized network.

**The shift from recorded to generated video is accelerating.** AI world models now create entire scenes frame-by-frame in real-time, responding instantly to user input. This technology is converging across: - Gaming environments that generate as players move - Robotics systems processing visual data - Creative tools with live AI effects - Interactive streaming platforms **The infrastructure challenge is significant.** Real-time AI video requires sub-100ms latency and 1000x more compute than text or image AI. Traditional GPU clouds struggle with these demands. **What's emerging:** - Personalized video adapting to each viewer - World model games generating environments on-the-fly - AI avatars requiring instant video processing - Dynamic content personalization at scale The gap between what's possible and existing infrastructure creates opportunities for new approaches to distributed computing.

**Real-time AI video processing** is reaching a new milestone: the ability to identify and replace products, billboards, and brands in video streams based on individual viewers. **How it works:** - AI scans video scenes in real-time - Identifies objects like coffee cups, billboards, products - Swaps in different brands depending on who's watching - Requires sub-second latency and significant GPU capacity This technology extends beyond advertising. Applications include: - Transforming webcam feeds instantly - Generating game environments that respond to player input - Live sports analysis during games The technical challenge: generating every frame on-the-fly with less than 100ms latency. Generic GPU clouds struggle to meet these demanding requirements for real-time processing.

Tencent is moving beyond traditional interruption-based advertising by **seamlessly embedding video ads directly into movies**. This represents a shift toward integrated, personalized advertising experiences. - The technology allows for real-time ad insertion within film content - Marks a departure from conventional ad breaks and pre-roll formats - Raises questions about infrastructure requirements for personalized content delivery The approach highlights limitations of existing centralized cloud systems, which weren't designed to handle this level of dynamic, personalized content modification at scale.