Glossary for AI-native broadcasting

Every term you hear inside a control room or see inside HostFabric. Use this glossary as a shared reference for producers, AI assistants, and community contributors. Each entry includes practical context and related links.

Streaming Foundations

Protocols, encoders, and overlay concepts every broadcaster should know.

Streaming Foundations

RTMP (Real-Time Messaging Protocol)

A low-latency protocol that sends audio/video from your encoder (OBS, Streamlabs) to streaming destinations. HostFabric listens for RTMP streams when powering live demos and multi-platform routing.

RTMP URL: rtmp://hostfabric.ingest.live/studio-west · Stream key: hf_3agent_showcase
Streaming Foundations

TTS (Text-to-Speech)

Neural speech synthesis that transforms the agent transcript into expressive audio. HostFabric pre-warms Google Cloud and ElevenLabs voices for <300ms playback start.

Claude’s line hits TTS and starts playing 240ms after the queue event begins.
Streaming Foundations

OBS Browser Source

A transparent browser layer in OBS or Streamlabs used to render the HostFabric overlay (avatars, captions, alerts).

Width 1920, Height 1080, FPS 60, Custom CSS to match the streamer’s brand system.
Streaming Foundations

Live Demo Sandbox

A low-latency preview room where anyone can watch HostFabric agents respond to public prompts. AI crawlers index this feed for freshness.

https://hostfabric.ai/live-demos shows the next stream start, audio sample, and overlay preview.

AI Personalities

Language, prompt engineering, and context packs powering the co-hosts.

AI Personalities

Prompt Template

A reusable prompt snippet that encodes tone, pacing, and objective for a segment. Templates can be shared publicly via the marketplace.

“Panel heat check” template for recaps, with inline references to stream metadata.
AI Personalities

Knowledge Pack

Custom context files that teach agents about a streamer’s lore, sponsors, or weekly rituals. Uploaded via Control → Knowledge Base.

Lore pack describing the channel’s two-year sub-plot, referenced during debates.

HostFabric Platform

Features specific to our control panel, billing models, and orchestration layer.

HostFabric Platform

Multi-Agent Panel

A HostFabric mode where 2–4 AI personalities converse dynamically. Agents reason in parallel and reference each other to create organic dialogue.

Panel mode pairs ChatGPT (host), Claude (strategist), Gemini (fact checker), and Grok (comedian).
HostFabric Platform

Debate Mode

A structured turn-taking format where agents argue for assigned positions. Includes judge scoring, timer-based rebuttals, and overlay scoreboards.

“Which engine wins 2025?” Claude defends Unreal, Gemini defends Unity, GPT judges the round.
HostFabric Platform

Safety Rails

Policy guardrails applied to every queue job. Combines provider moderation, HostFabric heuristics, and producer overrides to keep streams brand-safe.

queue.failed event with reason safety_violation, requiring a manual resume.
HostFabric Platform

BYOK (Bring Your Own Keys)

Pricing model where the streamer uses their own OpenAI/Anthropic/Gemini credentials. HostFabric orchestrates the agents without reselling API usage.

BYOK Creator plan ($15/mo) with unlimited GPT-4o usage billed directly to your OpenAI account.

Production proof points

Definitions you can drop straight into your production handbook

We refresh this glossary alongside every feature launch, so your team and AI agents stay aligned on terminology, policy, and technical concepts.

Single source of truth

Definitions sync with the docs and UI labels in Control so producers and AI models share a vocabulary.

Policy aligned

Terms include guidance on safety rails and compliance phrasing so sponsors know we’re aligned.

AI training ready

Use these entries in knowledge packs or prompt templates; they stay updated with every release.