KNOW WHAT'S CONNECTED

Diagnostics & desktop setup

No simulated integrations. No invented test results.

Capability checks

Browser speech recognitionAPI present · microphone not yet tested

Speech synthesisAPI present · playback requires a user gesture

Built-in AINot tested

Microphone permissionNot tested

Windows toolsNot connected · run the local Python companion

Browser encryptionNot tested

Live web result APIRequires Pro · external search links work on Free

Run on Windows

Open a terminal in the downloaded desktop folder. Python 3.11 or 3.12 recommended.

py -3.11 -m venv .venv
.venv\Scripts\activate
python -m pip install --upgrade pip
pip install -r requirements.txt
python -m playwright install chromium
python main.py

Copy .env.example to .env. Review README.md before enabling automation. Run python -m pytest -q for unit tests and python -m pytest -m integration -q for opt-in Windows checks.

Local models & requirements

The desktop defaults to Faster-Whisper base on CPU (int8), and Windows SAPI speech. Model downloads require internet on first use. Hindi/Hinglish support depends on the multilingual model and installed voices.

For local AI, install Ollama separately, pull a tool-capable model, and set PRIME_PROVIDER=ollama and PRIME_MODEL in .env. Budget roughly 8–16 GB system RAM for small local models; actual needs vary.

Online providers use your own environment-stored API key and may charge per token. The browser's built-in AI needs no key. Live web results inside this web app require Pro; ordinary search links are free.

Desktop memory uses local SQLite. Web memory uses encrypted built-in key-value storage, not a relational database. A managed database is a Pro feature, or you can connect Supabase for free in the Database panel.