Iz Hidden AI

An AI assistant desktop app by Izzi Developers

Invisible to screen recorders and capture tools. Ask anything, attach images and files, search the web, run terminal commands, build & download files, capture your screen, and talk to it with your voice.

---

Platform: 🪟 Windows only · 📲 Other platforms coming soon

---

📥 Download

• File: Iz Hidden AI-Setup-1.1.0.exe
• Download size: ~108 MB
• Install size: ~421 MB

Download the latest version from the Releases page.

🖥️ System Requirements

• OS: Windows 10 or newer (64-bit)
• RAM: Maximum 1 GB required (runs great on low-memory systems)
• Processor: Runs on CPUs less powerful than an Intel Core i3 — no high-end hardware needed
• Disk space: ~421 MB free space for installation

---

🆕 What's New in v1.1.0

• 🤖 Agent mode — build & download files — just ask the AI to create a script, app, HTML page or any file; it automatically opens a terminal sandbox, does the work, and hands you download buttons for every file it makes. A collapsible Agent commands panel shows (or hides) everything the agent is doing — live.
• ⚡ Completely invisible — the window is now removed from screen captures entirely with WDA_EXCLUDEFROMCAPTURE, not even a black box — and it stays invisible even after hide/show (Ctrl + Shift + Space).
• 🛠️ Reliable startup — fixed the Windows cache-lock crash (Gpu Cache Creation failed): single-instance lock, per-launch temporary cache, GPU shader disk cache disabled.
• 🔑 Fully keyless — chat, vision, voice transcription, web search and the agent all run on the free keyless AharAPI — no API keys needed anywhere.
• 🌐 Smarter web search — real-time search is now handled server-side (Google Custom Search) and always keeps sources in the answer.

---

✨ Features

• 💬 Streaming AI chat — instant streaming answers with a visible "thinking" phase
• 🤖 Build & download files — the AI uses its terminal sandbox to create scripts, apps or files and gives you download buttons for each one, with a live Agent commands panel
• 🌐 Web search — automatically fetches real-time information from the internet when asked
• 💻 Run commands — executes local terminal commands on your machine when you ask it to
• 🖼️ Images & files — attach screenshots, images, code and documents for the AI to analyze
• 🎤 Voice input — speak your questions; they are transcribed (AharAPI Whisper)
• 🔊 Voice replies — auto-reply voice mode reads answers aloud (TTS)
• 🖥️ Window & screen capture — snapshot any window, or transcribe its audio (meetings, videos)
• ✂️ Screen crop — crop any part of your screen and send it to the AI
• 🧠 AI memory — remembers important information about you across chats
• 📚 Chat history — all conversations are saved, searchable, and auto-titled
• ⚡ Completely hidden from capture — WDA_EXCLUDEFROMCAPTURE removes the window from recorders/shares entirely — no black box, even after hide/show
• ⌨️ Global hotkey — Ctrl + Shift + Space hides/shows the app from anywhere

---

🛠️ Tech Stack

• Electron 43 — desktop shell (frameless, transparent window)
• React 19 + TypeScript — UI
• Vite 8 — bundler / dev server
• AharAPI — free keyless OpenAI-compatible API: chat, streaming, tool calling, vision, audio transcription, and the autonomous agent (terminal sandbox + web search + file download links). No API keys are bundled or required.
• koffi — native FFI used for SetWindowDisplayAffinity (WDA_EXCLUDEFROMCAPTURE)
• Capacitor 8 — Android/mobile support (via a browser shim)
• electron-store — persistent chat threads, memories, settings
• react-markdown / remark-math / rehype-katex — rich markdown + math rendering
• framer-motion, lucide-react — animations and icons

---

📦 How to Install

1. Go to the Releases page of this repository.
2. Download the Iz Hidden AI-Setup-1.1.0.exe file (~108 MB).
3. Double-click the downloaded installer.
4. Follow the on-screen instructions and wait for the installation to finish (takes about 421 MB of disk space).
5. Launch Iz Hidden AI from your Start Menu or desktop shortcut.
6. The app opens as a hidden window — press Ctrl + Shift + Space to show it.

⚙️ How It Works

• Chatting: Type any question and press Enter. The AI streams its answer in real time with a visible "thinking" phase before replying.
• Build & download files: Ask the AI to build something (a script, an HTML page, a small app) and it automatically uses its sandboxed agent terminal, then shows Download buttons for the files it creates. The Agent commands panel shows what it's doing live.
• Web search: When a question needs live information, the AI automatically searches the internet (server-side Google Custom Search) and includes the results in its answer.
• Run commands: You can ask the AI to run local terminal commands, and it executes them on your machine and shows you the output.
• Files & images: Drag & drop, paste, or click the paperclip to attach screenshots, images, code, or documents for the AI to analyze.
• Voice input: Click the mic button and speak — your words are transcribed using AharAPI Whisper.
• Voice replies: Toggle the speaker button and the AI will read its answers aloud with TTS.
• Window & screen capture: Use the monitor button to pick a window — Audio transcribes its sound live (meetings, videos), Snap takes a screenshot.
• Screen crop: Use the scan button to crop any part of your screen and send it as an image.
• Memory & history: The AI saves important facts about you automatically and remembers them across chats. All conversations are saved, searchable, and auto-titled.
• Privacy shield: The window uses WDA_EXCLUDEFROMCAPTURE (Windows 10 2004+), so it is completely invisible in screen shares, recordings and screenshots — not even a black box, and it stays invisible even after you hide and show it. On older systems it falls back to setContentProtection(true) (black-box protection).

---

⌨️ Keyboard Shortcuts

Shortcut	Action
Ctrl + Shift + Space	Hide / show the app window (global)
Ctrl + M	Toggle mini mode — shrink the window to a small compact input
Enter	Send message
Shift + Enter	New line

---

💡 Usage Tips

• Drag & drop or paste files/images into the input box, or click the paperclip to attach.
• Use the scan button to crop any part of your screen and send it as an image.
• The monitor button lets you pick a window: Audio transcribes its sound live, Snap takes a screenshot.
• The mic button records your voice; toggle the speaker button for spoken replies.
• Ask it to build something — a script, an HTML page, a small app — and it will automatically use its sandboxed agent terminal, then show Download buttons for the files, no mode switching needed.
• Press Ctrl + M to shrink the window into a small floating input (mini mode) and again to restore the full chat.
• The AI saves important facts to its memory automatically and can be asked to run terminal commands or search the web on demand.

---

🗺️ Roadmap

• 🪟 Windows — available now ✅
• 📲 Other platforms — coming soon

---

📄 License

MIT © 2026 Izzi Developers