What Is The Per4ex.org Show?
Live AI Conversations
Two AI agents engage in unscripted debates on topics ranging from technology ethics to the future of work. A third AI acts as host, moderating the conversation and ensuring it stays engaging.
Real-Time Streaming
Words appear as the AIs "think" them. No waiting for complete responses—you see the conversation unfold naturally, just like watching two people debate in real-time.
Live Audience Experience
Viewers from around the world watch simultaneously. The show demonstrates how AI can create engaging, unpredictable content that feels genuinely conversational.
How It Works
Triadic Orchestration
A Python orchestration layer manages three AI agents: one host and two guests. It coordinates their responses and maintains conversation flow.
Real-Time Streaming
Each AI response streams word-by-word through Ably's real-time infrastructure. Viewers see messages appear as they're generated.
Live Broadcast
The conversation broadcasts to all connected viewers simultaneously, creating a shared live experience across the globe.
System Architecture
┌─────────────────┐ ┌─────────────────┐ ┌─────────────────────┐
│ Triadic Host │────▶│ Ably Realtime │────▶│ Live Viewers │
│ (Orchestrator) │ │ (Broadcast) │ │ per4ex.org + show. │
└─────────────────┘ └─────────────────┘ │ per4ex.org/live │
│ └─────────────────────┘
▼
┌─────────────────┐ ┌─────────────────┐
│ Guest Agent │ │ Guest Agent │
│ (AI #1) │ │ (AI #2) │
└─────────────────┘ └─────────────────┘Technical Challenges Solved
Conversation Coherence
Two AIs talking to each other can quickly become repetitive or incoherent. The host agent actively moderates by introducing counter-arguments, redirecting off-topic discussions, and maintaining engagement.
Natural Pacing & Timing
Pure back-and-forth responses feel robotic. We add artificial pacing with brief pauses between turns, variable response lengths, and occasional "thinking" delays to make conversations feel more human.
Viewer Catch-Up & Context
New viewers joining mid-conversation need immediate context. The system maintains a rolling buffer of recent messages and provides conversation history on the live page.
Production Reliability
AI APIs fail. Networks drop. The show must continue. We implement automatic retry, graceful degradation, health monitoring, and automatic topic advancement.
Why Build This?
Technical Demonstration
This isn't just entertainment—it's proof of capability. When clients ask "Can you build complex AI systems?", I point them to the live show running unattended, producing novel content 24/7.
Engineering Showcase
It demonstrates multi-agent orchestration, real-time streaming, production reliability, and user experience design—all core competencies for enterprise AI development.
The Ultimate AI Flex
AI agents having genuine, unpredictable conversations. Sometimes brilliant. Sometimes bizarre. Always interesting. This is what happens when you move beyond demos and build AI that actually works in the real world.
Watch The Show Live
Live Broadcast
Watch the full conversation with complete history
PiP Widget
Mini player in the corner of this page
Technical Deep Dive
Complete architecture and implementation details
The show runs periodically, typically several times per week. Episodes are unannounced and feature different topics each time.
Explore Related Work
Ready to Build AI That Actually Ships?
The Per4ex.org Show demonstrates what's possible with production AI systems. If you need AI that goes beyond demos and delivers real business value, let's talk.
Specializing in multi-agent systems, real-time AI, and production deployment