JAI - Your Personal AI Assistant

JAI (Just an Assistant, Intelligent) is a modular, networked personal assistant inspired by Jarvis. JAI combines polite, calm, and witty personality with powerful operational capabilities.

Why JAI

JAI is a self-hosted, local-first assistant that runs on your Windows PC via FastAPI. It emphasizes privacy, extensibility, and practical day-to-day usefulness.

• Local-first and private: Run the server on your own machine. Protect credentials in .env; no secrets in version control.
• Modular and extensible: Clear modules for personality, memory, TTS, controls, and APIs. Easy to add new intents and skills.
• Built-in memory system: Short-term and long-term memory (SQLite) with importance scoring and search.
• Multi-language out of the box: Translate responses to Hindi, Urdu, Arabic, Russian, and Spanish.
• Voice-ready: Text-to-Speech with configurable voice; voice commands on the roadmap.
• PC & media control: Open apps, lock screen, and more; media controls and YouTube integration are planned/available depending on your branch.
• Secure by default: HTTP Basic Auth for all commands; environment-based configuration for API keys.
• Observability: Commands are logged with user attribution for audits and debugging.

Better localization

• Translate responses into key languages without changing your workflow.
• Keep tone consistent via the personality module while adapting to user language preferences.
• Run locally to respect regional data residency preferences.

Superior privacy & control

• Host the assistant on your machine or LAN.
• Use Basic Auth and store keys in .env.
• Keep long-term memory local in SQLite; logs provide a clear audit trail.

Business Benefits

JAI can drive measurable outcomes across operations, support, and revenue enablement.

• Cost reduction

  • Consolidate multiple assistant/SaaS tools into one local service.
  • Reduce external API usage by caching and only calling what you need.
  • Simpler IT surface area (single venv + .env), fewer vendor lock-ins.
  • Example KPIs to track: monthly API spend, number of SaaS seats replaced, average time per routine task.

• Automation

  • Automate repetitive desktop tasks (open apps, run commands), information pulls (weather, news), and daily summaries.
  • Expand with intents for reminders, media control, and more as needed.
  • Example KPIs: tasks automated per week, minutes saved per user, number of manual steps removed.

• Customer service improvement

  • Multilingual responses (Hindi/Urdu/Arabic/Russian/Spanish) for internal helpdesks or branches.
  • Consistent, polite tone via the personality module; quick responses via HTTP endpoints.
  • Example KPIs: first-response time, CSAT, average resolution time.

• Sales growth enablement

  • Draft outreach snippets, summarize news, and generate spoken prompts with TTS.
  • Assist with research and content prep; integrate lightweightly with desktop workflows.
  • Example KPIs: meeting booking rate, opportunity creation rate, time-to-first-touch.

Business Impact Examples

• Cost reduction

  • Consolidate overlapping SaaS assistants into JAI and reduce external API calls via caching and selective requests.
  • Example: 10 users × 2 tools at $15/user/mo ≈ $300/mo saved; API optimizations cut $300 → $150–$210/mo.
  • Track: monthly API spend, SaaS seats replaced, cost per automated task.

• Automation

  • Automate routine desktop tasks (open apps, run commands), info pulls (weather/news), and daily summaries.
  • Example: 8 tasks/day × 1.5 min × 10 users ≈ 120 min/day → ~40 h/month. At $20/h ≈ $800/mo saved.
  • Track: tasks automated/week, minutes saved/user, manual steps removed.

• Customer service improvement

  • Multilingual responses (Hindi/Urdu/Arabic/Russian/Spanish) and consistent tone improve speed and quality.
  • Example: First-response time 15 → 5 min; CSAT +5–10 pts with quicker, clearer answers.
  • Track: FRT, CSAT, average resolution time.

• Sales growth enablement

  • Draft outreach snippets, summarize news, and generate spoken prompts with TTS to speed prep.
  • Example: +5% meetings booked (e.g., 20 → 21/mo). With 20% win rate and $500 margin/deal ≈ +$100 MRR.
  • Track: meeting booking rate, opportunity creation rate, time-to-first-touch.

Features

Core Functions ✓

• Weather Updates: Real-time weather data with caching
• News Headlines: Top news from NewsAPI
• Memory System: Short-term and long-term context awareness
• Multi-language Support: Translate responses to Hindi, Urdu, Arabic, Russian, Spanish
• Text-to-Speech: Optional spoken responses with voice customization
• PC Control: Open apps, lock screen, and more (admin-restricted)
• Networked API: FastAPI server with HTTP Basic Auth

Advanced Features (In Progress)

• Voice Commands: Speech recognition with wake word detection
• Music & YouTube Control: Play, pause, search, and control media
• Calendar & Reminders: Schedule events with alerts
• Windows Startup Service: Auto-launch on boot
• Custom Voice Packs: Personalize JAI's tone

Personality

JAI is designed to be:

• Polite: Professional and respectful
• Calm: Measured responses, never rushed
• Witty: Occasional humor and charm

Installation

1. Clone or Download

cd "C:\Users\Abdul Rahman\Documents\JAI_Assistant"

2. Create Virtual Environment

python -m venv venv
.\venv\Scripts\Activate.ps1

3. Install Dependencies

pip install -r requirements.txt

4. Configure Environment

Create a .env file in the project root:

# Required
OPENAI_API_KEY=your_openrouter_api_key_here

# Optional
OPENWEATHER_API_KEY=your_weather_api_key
NEWS_API_KEY=your_newsapi_key

# Speech (set to true to enable spoken responses)
SPEAK_RESPONSES=false
TTS_VOICE=David

Get API Keys:

• OpenAI (via OpenRouter): https://openrouter.ai/
• Weather: https://openweathermap.org/api
• News: https://newsapi.org/

Usage

Start the Server

python jai_assistant.py

Server runs on http://0.0.0.0:8000

Send Commands (HTTP)

curl -u user1:pass1 -X POST http://localhost:8000/command `
  -H "Content-Type: application/json" `
  -d '{"command": "weather in Dubai"}'

Example Commands

• "weather in Canada"
• "news"
• "my name is Abdul"
• "remember my favorite color is blue"
• "recall favorite color"
• "open calculator"
• "lock screen" (admin only)
• "speak hindi" (switch to Hindi responses)

Architecture

JAI_Assistant/
├── jai_assistant.py       # Main FastAPI server & command executor
├── personality.py         # Personality prompt builder & quips
├── tts.py                 # Text-to-speech module
├── memory.py              # Short/long-term memory (SQLite)
├── jai_controls.py        # PC control commands
├── requirements.txt       # Python dependencies
├── .env                   # Configuration (API keys, settings)
└── README.md              # This file

Modules

personality.py

Centralizes JAI's tone and prompt generation. Ensures consistent polite, calm, witty responses.

tts.py

Safe text-to-speech with:

• Voice selection (male/female, by name)
• Timeout protection
• Non-blocking operation

memory.py

SQLite-backed memory:

• Short-term: Recent conversation context
• Long-term: User preferences, facts (with importance scoring)

jai_controls.py

System control commands:

• Open applications (Notepad, Calculator, Chrome, etc.)
• Future: Volume, media playback, shutdown/restart

Security

• HTTP Basic Auth: Username/password required for all commands
• Admin Restrictions: Sensitive commands (shutdown, lock) require admin role
• API Key Protection: Store keys in .env, never commit to version control
• Logging: All commands logged with user attribution

Roadmap

High Priority

• Speech-to-Text (STT) module with wake word
• Media controls (volume, play/pause, YouTube)
• Calendar & reminders with APScheduler
• Windows startup service

Medium Priority

• Plugin system for extensibility
• Enhanced memory with semantic search
• Multi-device sync (cloud storage)
• Web dashboard UI

Future Extensions

• IoT & smart home integration
• Secure password manager
• Finance advisor plugin
• Study helper & coding assistant plugins

Troubleshooting

TTS Not Working

• Ensure pyttsx3 is installed: pip install pyttsx3
• Check available voices: Run test_tts.py
• Set SPEAK_RESPONSES=true in .env

Speech Recognition Issues

• Install PyAudio: pip install pyaudio (may require Microsoft C++ Build Tools)
• Test microphone: Run test_speech.py

API Errors

• Verify API keys in .env
• Check internet connection
• Review jai_assistant.log for details

Contributing

JAI is designed to be modular and extensible. To add features:

1. Create a new module (e.g., jai_media.py)
2. Add intent patterns in classify_intent()
3. Handle the intent in execute_command()
4. Update requirements.txt if new dependencies are needed

License

Personal project - feel free to adapt and extend for your own use.

Credits

Built with:

• OpenAI GPT-4o-mini (via OpenRouter)
• FastAPI
• pyttsx3
• SpeechRecognition
• deep-translator

---

JAI: "Just for you, I've polished my circuits to shine!"