XEXAMAI

Your smart free assistant for interviews and exams

⭐ Star this repository if it helped you! ⭐

Table of Contents

• Key Features
• How to Use
• How to Use Locally
  • Local LLM Processing
  • Local Speech Recognition
• Important Notes
• For Developers

🚀 Key Features

• FREE USAGE - no subscription required, no limits for local
• Complete invisibility - stays hidden during screen sharing in Zoom, Google Meet, Teams and other platforms
• Advanced AI models - uses cutting-edge speech recognition and AI generation
• Local AI models - use local AI models for faster processing
• Privacy & Security - all data processed locally, audio is not stored
• Flexible audio settings - choose between system sound and microphone
• Cross-platform - works on Windows, macOS and Linux
• Simple interface - intuitive and easy to use
• Customizable transcription - choose from multiple AI models and customize prompts

If you have any issues using the app, please open an issue

🎯 How to Use

1. Setup

1. Open xexamai application
2. Go to Settings tab
3. Enter api key:
   • OpenAI API key (get it from platform.openai.com)
   • Google AI API key (get it from console.cloud.google.com)
4. Choose one LLM Model of the models suitable for your key
5. Choose audio input device:
   • System Audio - for recording sound from applications (Zoom, Teams, etc.)
   • Microphone - for recording your voice
6. Configure transcription settings:
   • Transcription Model - choose from:
     • Whisper-1 (Default) - balanced speed and accuracy
     • GPT-4o Transcribe (High Quality) - maximum accuracy for complex audio
     • GPT-4o Mini Transcribe (Fast) - optimized for speed and efficiency
   • Transcription Prompt - customize how AI should process your audio:
     • Default prompt optimized for technical interviews in Russian
     • Preserves English technical terms (Redis, Postgres, API, etc.)
     • Can be customized for different languages and contexts

2. Usage

1. Switch to Main tab
2. Click Start Audio Loop - the app will start recording audio in the background
3. When needed, click Send Last X Seconds to get an AI response
4. Get instant answers to help you during interviews or exams
5. If you use Google AI, then you can switch Transcription Type to Stream and then Gemeni will show you a recognized text that you can send to receive an answer at the right time.
6. See what settings are in different sections (Hotkeys, API timeout for retry, always on top, opacity)

3. Usage Tips

• For interviews: use system audio to record interviewer's questions
• For exams: configure microphone to record your questions
• Adjust transparency for maximum stealth
• Practice before important events

Support the project on https://pump.fun/coin/D1zY7HRVE4cz2TctSrckwBKnUzhCkitUekgTf6bhXsTG or contact with me for other creds

How to Use Locally

The examples below are implemented and tested on Windows 11. Steps may differ on other systems.

Local streaming as in gemeni is not available, but this is not scary

The assistant works in two stages:

1. Audio transcription

2. Getting an answer from the LLM

Each stage can be run locally.

1. Install CUDA

   https://developer.nvidia.com/cuda-12-1-0-download-archive?target_os=Windows&target_arch=x86_64&target_version=11&target_type=exe_local

2. cuDNN 9.13.1

   https://developer.nvidia.com/cudnn-downloads?target_os=Windows&target_arch=x86_64&target_version=11&target_type=exe_local

3. Add to Windows PATH environment variable

   C:\Program Files\NVIDIA\CUDNN\v9.13\bin\12.9

   Or the path where cuDNN was installed

4. Restart the PC

Local LLM Processing

Minimum recommended configuration:

• CPU - 4 cores / 8 threads
• GPU - 6 GB VRAM
• RAM - 16 GB

1. • In xexamai settings select an Mode -> LLM = Local.
   • In xexamai settings choose a Model -> LLM from the available models:
     • gpt-oss:120b gpt-oss:20b gemma3:27b gemma3:12b gemma3:4b gemma3:1b deepseek-r1:8b qwen3-coder:30b qwen3:30b qwen3:8b qwen3:4b

       Choose a smaller model if your PC is low-spec

2. Install Ollama

   https://ollama.com/

3. (Optional) Change the default models location

   • Remove the original models directory Remove-Item -Recurse -Force "C:\\Users\\xl\\.ollama\\models"
   • Then create a junction New-Item -ItemType Junction -Path "C:\\Users\\xl\\.ollama\\models" -Target "F:\\ollama_models\\models"

4. Download the model chosen earlier

   ollama pull qwen3:4b

5. Start Ollama

   ollama serve

Local Speech Recognition

1. In xexamai settings select an Mode -> Transcription = Local.

2. In xexamai settings choose one of Model -> Transcription

3. In xexamai settings choose Local transcription device: GPU (Graphics/NVIDIA) or CPU (Processor)

4. Install and run fast-fast-whisper

   It does not auto-start with Windows; you need to launch it manually so the local speech recognition server is running

The first use after the opening of the program will be slower, since with local use of the AI models will be loaded in GPU or RAM, which takes time. Before the interview, do the 1st question and get the answer so that the subsequent calls are faster.

Important Notes

• This application is intended for educational purposes
• Ensure that AI assistance is allowed in your situation
• Follow honesty and academic integrity rules
• API keys are stored locally and not shared with third parties

🔧 For Developers

Contributing

We welcome contributions to the project! If you want to contribute:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

Local Development

Requirements

• Node.js 20+
• npm or yarn

Installation

# Clone the repository
git clone https://github.com/Artasov/xexamai.git
cd xexamai

# Install dependencies
npm install

# Build the project
npm run build

# Run in development mode
npm run dev

Project Structure

src/
├── main/           # Electron main process
├── renderer/       # Renderer process (UI)
├── preload/        # Preload scripts
└── shared/         # Shared types and utilities

Available Commands

• npm run dev - run in development mode
• npm run build - build the project
• npm run build:win - create portable Windows executable
• npm run build:mac - create portable macOS executable (macOS only)
• npm run build:linux - create portable Linux directory
• npm run build:all - create portable executables for Windows and Linux
• npm run build:win-linux - same as build:all
• npm run clean - clean build directory

Building for Different Platforms

Windows

npm run build:win

Creates:

• Portable executable (xexamai-${version}.exe)

macOS

npm run build:mac

Creates:

• ZIP archive for Intel and Apple Silicon (xexamai-${version}-x64.zip, xexamai-${version}-arm64.zip)

Note: For macOS builds, you may need to:

1. Install Xcode Command Line Tools: xcode-select --install

Linux

npm run build:linux

Creates:

• Portable directory (linux-unpacked/)
• Archive is automatically created (xexamai-${version}-linux-x64.tar.gz)

Note:

• Archive is ready for distribution - users can extract and run
• Building AppImage requires additional tools that are difficult to configure on Windows

Cross-platform Building

⚠️ Cross-platform build limitations:

• Windows: Can only build for Windows and Linux (via WSL)
• macOS: Can build for all platforms (Windows, macOS, Linux)
• Linux: Can only build for Linux and Windows, but not macOS

Cross-platform build commands:

# On Windows - only Windows and Linux
npm run build:all

# On macOS - all platforms
npm run build:all
npm run build:win
npm run build:mac  
npm run build:linux

# On Linux - only Linux and Windows
npm run build:all

To build macOS version:

• Use a macOS machine
• Or use GitHub Actions (if you set up CI/CD)

Technologies

• Electron - cross-platform desktop application
• TypeScript - typed JavaScript
• Tailwind CSS - utility-first CSS framework
• OpenAI API - AI integration

---

Made with ❤️ for successful interviews and exams