NEWsightAI

A local Python assistant that can:

1. Watch a selected monitor (screen capture).
2. Let you ask by voice or text about what is on that screen.
3. Answer using a vision-capable AI model.
4. Optionally speak responses aloud.

Requirements

• Python 3.10+
• An OpenAI API key (OPENAI_API_KEY)
• Microphone access
• Desktop audio support for text-to-speech

Setup

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Set your API key:

export OPENAI_API_KEY="your_api_key_here"

Usage

List monitors first:

python app.py --list-monitors

Run with monitor 1:

python app.py --monitor 1

Enable spoken AI replies:

python app.py --monitor 1 --voice-reply

Interaction

• Type a question and press Enter, or
• Press Enter on an empty line to speak through your microphone.
• Type /quit to exit.

Notes

• If the AI cannot read details, ask it to describe larger UI areas or zoom your screen.
• PyAudio installation can require OS-level packages (for example portaudio).
• Change model with --model if you prefer a different vision-capable OpenAI model.