NeuralAgent is your AI personal assistant that actually gets things done. It lives on your desktop, types, clicks, navigates the browser, fills out forms, sends emails, and performs tasks automatically using modern large language models — all powered by a fast, extensible, and open architecture.

Real productivity. Not just conversation.

• 🌍 Website: https://www.getneuralagent.com
• 💬 Discord: Join NeuralAgent Discord

• ✅ Desktop automation with pyautogui
• ✅ Background automation (Windows Only For Now) via WSL (browser-only).
• ✅ Supports Claude, GPT-4, Azure OpenAI, and Bedrock
• ✅ Modular agents: Planner, Classifier, Suggestor, Title, and more
• ✅ Multimodal (text + vision)
• ✅ FastAPI backend + Electron + React frontend

neuralagent/
├── backend/              # FastAPI + Postgres backend
├── desktop/              # ElectronJS desktop app
│   └── neuralagent-app/  # React frontend inside Electron
│   └── aiagent/          # Python code (pyautogui)
└── README.md

🧪 Open two terminal windows – one for backend and one for desktop.

1. Create and activate a virtual environment (optional but recommended):

cd backend
python -m venv venv
# Activate:
source venv/bin/activate  # macOS/Linux
venv\Scripts\activate     # Windows

2. Install requirements:

pip install -r requirements.txt

3. Create a local Postgres database.

4. Copy .env.example to .env and fill in:

DB_HOST=
DB_PORT=
DB_DATABASE=
DB_USERNAME=
DB_PASSWORD=
DB_CONNECTION_STRING=

JWT_ISS=NeuralAgentBackend
JWT_SECRET=

REDIS_CONNECTION=

# Optional: For Bedrock
AWS_ACCESS_KEY_ID=
AWS_SECRET_ACCESS_KEY=
BEDROCK_REGION=us-west-2

# Optional: For Azure OpenAI
AZURE_OPENAI_ENDPOINT=
AZURE_OPENAI_API_KEY=
OPENAI_API_VERSION=2024-12-01-preview

# Optional: OpenAI/Anthropic
OPENAI_API_KEY=
ANTHROPIC_API_KEY=

# Model config per agent
CLASSIFIER_AGENT_MODEL_TYPE=openai|azure_openai|anthropic|bedrock
CLASSIFIER_AGENT_MODEL_ID=gpt-4.1

TITLE_AGENT_MODEL_TYPE=openai|azure_openai|anthropic|bedrock
TITLE_AGENT_MODEL_ID=gpt-4.1-nano

SUGGESTOR_AGENT_MODEL_TYPE=openai|azure_openai|anthropic|bedrock
SUGGESTOR_AGENT_MODEL_ID=gpt-4.1-mini

PLANNER_AGENT_MODEL_TYPE=openai|azure_openai|anthropic|bedrock
PLANNER_AGENT_MODEL_ID=gpt-4.1

COMPUTER_USE_AGENT_MODEL_TYPE=openai|azure_openai|anthropic|bedrock
COMPUTER_USE_AGENT_MODEL_ID=us.anthropic.claude-sonnet-4-20250514-v1:0

LANGCHAIN_TRACING_V2=false
LANGCHAIN_ENDPOINT=
LANGCHAIN_API_KEY=
LANGCHAIN_PROJECT=

# Google Login
GOOGLE_LOGIN_CLIENT_ID=
GOOGLE_LOGIN_CLIENT_SECRET=
GOOGLE_LOGIN_DESKTOP_REDIRECT_URI=http://127.0.0.1:36478

5. Run database migrations:

alembic upgrade head

6. Start the backend server:

uvicorn main:app --reload --host 0.0.0.0 --port 8000

1. Install dependencies in the Electron root:

cd desktop
npm install

2. Navigate to the React app:

cd neuralagent-app

3. Copy .env.example to .env and fill in:

REACT_APP_PROTOCOL=http
REACT_APP_WEBSOCKET_PROTOCOL=ws
REACT_APP_DNS=127.0.0.1:8000
REACT_APP_API_KEY=

4. Go back to the desktop root:

cd ..

5. Set up the local AI agent daemon (Python service):

cd aiagent
python -m venv venv
source venv/bin/activate  # Or use `venv\Scripts\activate` on Windows
pip install -r requirements.txt
deactivate

6. Start the Electron desktop app:

cd ..
npm start

You can configure different model providers (OpenAI, Azure OpenAI, Anthropic, Bedrock) per agent in .env.
Agent types include:

• PLANNER_AGENT
• CLASSIFIER_AGENT
• TITLE_AGENT
• SUGGESTOR_AGENT
• COMPUTER_USE_AGENT

We welcome pull requests and community contributions!

MIT License.
Use at your own risk. This tool moves your mouse and types on your behalf — test responsibly!

Feel free to open an issue or start a discussion.