Skip to content

Getting Started

CRAFT Studio is a YouTube content creation platform with AI-powered idea generation, script writing, video production, audio, competitive research, and multi-source resource discovery — backed by PostgreSQL, NATS workers, and 14 specialized AI agents.

What You Can Do

  • Generate Ideas — AI-powered brainstorming with channel personality, or save from YouTube research
  • Write Scripts — AI write, revise, polish with fact-checking and humanization (Claude, Gemini, Ollama)
  • Produce Videos — 65+ Remotion components with an 8-stage pipeline orchestrated by AI agents
  • Research Competitors — YouTube search with outlier detection, channel deep dives, and earnings estimates
  • Curate Content — AI-scored proposals, approval workflows, and episode pipeline tracking
  • Produce Audio — 300+ TTS voices with screenplay-aware parsing, GPU speech synthesis, and voice cloning
  • Find Resources — Search 14 royalty-free sources for video, image, audio, and reference material
  • Track Jobs — Async NATS workers for AI generation, rendering, and downloads with real-time SSE updates
  • Go Mobile — Android companion app with full dashboard parity over your local network

Layout

The Studio is organised around a top chrome — a single header bar — rather than a left sidebar. In it:

  • Logo · channel pill · peer avatars on the left
  • Stage rail across the middle (Discover → Ideas → Scripts → Audio → Episodes) — the five primary workflow stages
  • ⌘K command palette slot in the middle-right, then Resources · Settings · Marketplace · Feedback on the far right
  • Avatar menu top-right (API keys, password change, theme, sign out)

Channel-scoped actions (Proposals, Jobs, channel Settings) live inside the channel-pill dropdown. See Top chrome & ⌘K for the full tour.

Architecture

Browser ──> Studio API ──> Studio runtime
              |-- AI (Claude Agent SDK + Gemini + Ollama)
              |-- 14 AI Agents (research, write, storyboard, composite...)
              |-- YouTube research (yt-dlp + Data API)
              |-- TTS (Edge TTS, ElevenLabs, OpenAI, OpenedAI Speech)
              |-- Background workers (idea, tts, discover, render...)
              |-- Role-based multi-tenant access
              |-- Feedback → GitHub Issues (AI grooming + worker)
              |-- Persistent storage + caching
              |-- Remotion Video (65+ components, Chromium render)
              \-- GPU Services (Ollama, ComfyUI, MusicGen)

Mobile App ──> Same Studio API (local network)

Getting into the Studio

Once you're signed in, the top chrome gives you everything:

  • Channel pill on the left — switch channels, or create your first one.
  • Stage rail across the middle — Discover → Ideas → Scripts → Audio → Episodes.
  • ⌘K — search and navigate everything without leaving the keyboard.

TIP

Only a Gemini or Claude API key (set in the avatar menu → API keys) is needed to start generating. Most features work without any paid keys at all.

Next steps