synapse-assistant-ai 0.2.6

Free offline voice-controlled AI assistant. Install → Run → Talk.

🤖 Synapse-Assistant-AI — Free Offline Voice-Controlled Assistant

Install. Run. Talk. That's it.

Synapse-Assistant-AI is a completely offline, privacy-first, voice-controlled AI assistant for your computer. It listens to your voice commands, processes them locally without sending your data to any cloud, performs desktop automation tasks, and speaks back to you.

---

🌟 Key Features

• 100% Offline & Private: Uses local LLMs (via Ollama) and offline Speech-to-Text (Vosk). No API keys, no internet required (except for initial setup and web searches).
• Voice-Controlled: Always listening for its wake word (e.g., "Jarvis", "Friday", or your custom name).
• Conversational Chat (v0.2.6): It's not just a strict bot! You can talk to it freely (e.g., "Write a song for me", "Tell me a joke", "Solve this math problem") and it replies naturally using its LLM brain.
• Smart Clarifications (v0.2.6): If you ask it to perform a task but forget to give details (like "Send a WhatsApp message" without a phone number), it will politely ask you for the missing information before executing.
• Context Memory: Remembers your recent conversation history so you can ask follow-up questions naturally.
• Vision & Clipboard: Can read your screen (OCR) and read your copied clipboard text.
• Plugin System: Easily extend the AI's capabilities by dropping Python scripts into a plugins folder.
• Desktop Automation: Open websites, launch apps, play media, and search the web hands-free.
• Smart Notifications: Automatically notifies you via a Telegram Bot when a background task completes.
• WhatsApp Integration: Automates sending WhatsApp messages to your contacts.

---

🚀 Installation

Ensure you have Python 3.10 or higher installed.

For Windows:

pip install synapse-assistant-ai

For macOS / Linux:

pip3 install synapse-assistant-ai

(Note: On macOS and Linux, you might need to use pip3 instead of pip depending on your setup).

---

⚙️ Quick Start & Setup

Run the assistant from your terminal:

synapse-ai

(Note: If synapse-ai says "command not found", scroll down to the Troubleshooting section below to fix it permanently!)

First-Time Interactive Setup: The first time you run the command, an interactive setup will start right in your terminal:

1. Name your AI: Choose a wake word (e.g., Jarvis, Friday, or Hey AI).
2. Telegram Notifications (Optional): If you want to receive task completion alerts on your phone, you will be prompted to paste your Telegram Bot Token and Chat ID directly in the terminal. No .env editing required!
3. Model Download: It will auto-download the necessary AI voice models (~1.1 GB, one-time only).

After that, it starts instantly. Just say your wake word:

"Jarvis, open youtube"

---

🗣️ What it can do

Here are some examples of what you can ask your assistant to do:

Say this	It does this
"Jarvis, open google"	Opens google.com in your default browser
"Jarvis, open youtube"	Opens youtube.com
"Jarvis, launch Safari"	Opens the Safari application (Mac)
"Jarvis, read my screen"	Takes a screenshot and reads the text visible on your screen (OCR)
"Jarvis, what's on my clipboard"	Reads the text you currently have copied
"Jarvis, play /path/to/song.mp3"	Plays the specified media file locally
"Jarvis, send whatsapp to [phone] — hello"	Automates typing and sending a WhatsApp message
"Jarvis, search web for latest AI news"	Opens Chrome & gives you a spoken summary of the top result
"Jarvis, send message — meeting at 5"	Sends a text message to your configured Telegram chat
"Jarvis, send voice note — I'll be late"	Records and sends a voice note to you on Telegram
"Jarvis, set reminder 5 minutes — drink water"	Sets a local timer and reminds you in 5 minutes
"Jarvis, what is the time?"	Instantly tells you the current local time
"Jarvis, can you write a short poem?"	Replies naturally with a generated poem using its LLM brain
"Jarvis, what is Python?"	Consults its local AI Brain (Ollama) and answers your question

---

🔌 Custom Plugins

Want to add your own features? It's incredibly easy. Just drop a Python script into ~/.superai/plugins/ that registers a new function. The AI will automatically read your plugin, learn what it does, and start using it the next time you talk to it!

---

💻 Requirements

To run Synapse-Assistant-AI smoothly, your system needs:

• OS: Windows, macOS, or Linux
• Python: Version 3.10 or higher
• Hardware:
  • Working Microphone
  • 6 GB RAM (minimum, 8GB+ recommended)
  • ~1.5 GB free disk space (for AI models)
• Internet: Only required for the first-time model download and web search commands.
• Screen Reading (OCR): To use the "read screen" feature, you must have Tesseract OCR installed on your system (brew install tesseract on Mac, or sudo apt install tesseract-ocr on Ubuntu).

---

🛠️ Troubleshooting

• Installation Errors (vosk / sounddevice): If you get a "No matching distribution found" error during pip install, it means you are using an experimental or unsupported version of Python (like Python 3.13 or 3.14). Please install a stable release like Python 3.12 and run python3.12 -m pip install synapse-assistant-ai.

• Command Not Found Error: If your terminal says command not found: synapse-ai, it means your system's PATH is not configured for Python CLI tools. We have built an automated setup to fix this!

  For Windows Users:

  1. Open your Command Prompt or PowerShell.
  2. Run the auto-setup command: python -m super_ai setup
  3. Close and reopen your Command Prompt. You can now use the synapse-ai command forever!

  For macOS / Linux Users:

  1. Open your Terminal.
  2. Run the auto-setup command: python3 -m super_ai setup
  3. Restart your terminal or run source ~/.zshrc (or source ~/.bashrc on Linux). You can now use the synapse-ai command flawlessly!

  (Fallback: If you still face issues, you can always run the assistant directly from anywhere by typing python3 -m super_ai or python -m super_ai)

---

🧠 How it works under the hood

You speak → STT Engine (Vosk) → AI Brain (Ollama/Llama 3) → Action / Answer → TTS Engine → Speaks back

Everything runs locally on your machine.

• No cloud servers.
• No recurring API bills.
• No data leaves your computer.

---

📄 License

This project is licensed under the MIT License.