> ## Documentation Index
> Fetch the complete documentation index at: https://docs.origami.techmitten.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Introduction to Origami AI — Browser AI Video Studio

> Learn how Origami AI uses WebGPU, WebLLM, Kokoro.js, and FFmpeg.wasm to run a full AI video studio in your browser with complete privacy.

Origami AI is a full-featured AI studio that runs entirely in your browser. Whether you're turning a presentation deck into a narrated MP4, producing vertical short-form content from a single prompt, or chatting with a local AI model, everything happens on your device. Unlike cloud-based video tools, Origami AI uses **WebGPU** to run large language models locally, so your content never leaves your machine — and no account or subscription is required to get started.

## What You Can Build

<CardGroup cols={2}>
  <Card title="PDF to Video" icon="file-pdf" href="/features/pdf-to-video">
    Upload a presentation deck, generate AI narration scripts per slide, add text-to-speech voiceovers and background music, then export a polished MP4 — all inside your browser.
  </Card>

  <Card title="Shorts Generator" icon="film" href="/features/shorts-generator">
    Describe a topic and Origami AI writes a scene-by-scene script, generates visuals, synthesizes a voiceover, and burns in captions for a complete vertical short in minutes.
  </Card>

  <Card title="Screen Recording" icon="circle-dot" href="/features/screen-recording">
    Record any browser tab or window with cinematic auto-zoom. Origami AI tracks your cursor interactions and applies smooth camera moves during idle periods automatically.
  </Card>

  <Card title="AI Assistant" icon="message-bot" href="/features/ai-assistant">
    Chat with local WebLLM models directly in your browser. Attach images or video clips for vision analysis and get instant answers — no internet connection needed after setup.
  </Card>

  <Card title="Voice Studio" icon="microphone" href="/features/voice-studio">
    Generate high-quality TTS audio from any text using Kokoro neural voices. Control speed, select from multiple voices, and export at 128, 192, or 320 kbps MP3.
  </Card>

  <Card title="File Studio" icon="folder-open" href="/features/file-studio">
    Convert images between formats and compress audio files entirely in your browser — no upload to any server required.
  </Card>
</CardGroup>

## Privacy First

Your data stays on your device. Origami AI stores all API keys and project data in your **browser's IndexedDB** — a local, sandboxed storage that is never transmitted to any server. When you use a cloud AI provider such as Gemini or an OpenAI-compatible endpoint, your credentials live only in your browser and are sent directly to that provider, bypassing Origami AI's servers entirely.

Local AI models run fully on-device via WebGPU. No slides, scripts, audio, or generated video frames are uploaded anywhere during processing. You can render a complete video with zero outbound network traffic after the initial model download.

## How AI Works in Origami

Origami AI combines four browser-native technologies to deliver a studio-quality experience without a server-side render farm:

* **WebGPU** — the modern browser graphics API that gives JavaScript direct, low-overhead access to your GPU. Origami AI uses it to accelerate both AI inference and video effects.
* **WebLLM** — an open-source runtime (`@mlc-ai/web-llm`) that compiles large language models to run on WebGPU. It powers narration script generation, the AI Assistant, Shorts scripting, and scene analysis.
* **Kokoro.js** — an in-browser neural text-to-speech engine that synthesizes natural-sounding speech audio for each slide or scene without any server round-trip. Choose from multiple voices and quantization qualities (q4 for speed, q8 for high quality).
* **FFmpeg.wasm** — the industry-standard FFmpeg video toolkit compiled to WebAssembly. It composites your slides, audio tracks, music, and pan/zoom effects into a 720p or 1080p MP4 directly on your device.

<Note>
  Local AI features require a **WebGPU-capable browser** (Chrome/Edge 113+, Firefox Nightly with `dom.webgpu.enabled`, or Safari 18+ on macOS Sonoma). If your browser doesn't support WebGPU, you can still use all other Origami AI tools and connect a cloud AI provider — such as Gemini, OpenAI, or any OpenAI-compatible API — via **⚙️ Settings → API** as a fallback for script generation and the AI Assistant.
</Note>