Let AI drive your computer — Autouse AI — Computer Use, now with both the macOS and Windows builds combined in a single repository. Control your entire OS with natural language. Browser automation, coding tasks, file management — all powered by vision-language models.
Features • Example Tasks • Providers • Setup • Author
- GUI-based automation
- Multi-agent system
- Web agent
- CLI agent
- AppScript
- Secure sandbox
Just describe what you want — Auto Use picks the right tool for the job.
"Open Chrome, go to YouTube, and search for Python tutorials"
"Create a Python Flask API with user authentication"
"Find the latest NVIDIA stock price and quarterly revenue"
"Check disk space and clean up temp files"
"Send an iMessage to John saying I'll be 10 minutes late"
| Category | Examples |
|---|---|
| Browser | Fill forms, extract data, navigate sites, download files |
| Productivity | Create documents, manage spreadsheets, organize files |
| Development | Write code, debug errors, run tests, manage git |
| System | Install software, configure settings, manage processes |
| Research | Search web, compile information, generate reports |
Auto Use supports 6 LLM providers:
- Anthropic
- Groq
- OpenAI
- OpenRouter
- Perplexity
- macOS (Apple Silicon or Intel) or Windows 10/11
- API Key from any supported provider
💡 Recommended for most users: We strongly encourage installing the latest binary build from our official website for a fully seamless installation and the complete UI experience — no manual setup required.
The steps below are for developers who want to run Auto Use directly from source.
-
Run the setup script
bash MacOS_setup.sh
-
Add your API key(s)
Copy the example env file and fill in your keys:
cp .env.example .env
Then open
.envand add the API key for whichever provider(s) you want to use. -
Run Auto Use
python main.py
-
Run the setup script
windows_setup.bat
-
Add your API key(s)
Copy the example env file and fill in your keys:
copy .env.example .envThen open
.envand add the API key for whichever provider(s) you want to use. -
Run Auto Use
python main.py
Note: On Windows, Python 3.13.3 is the preferred version for best compatibility.
- Sandbox Isolation — Code runs in a protected environment
- No System Modification — Won't delete files or run destructive commands without permission
- Permission Awareness — Asks for confirmation before accepting elevation prompts
- Path Protection — Blocks access to critical system folders
| Feature | Auto Use | Others |
|---|---|---|
| Multi-agent system | ✅ | ❌ |
| Domain knowledge injection | ✅ | ❌ |
| Multi-provider LLM support | ✅ | Limited |
| Vision-based automation | ✅ | ✅ |
| Coding agent | ✅ | ❌ |
| Web search integration | ✅ | ❌ |
| Secure sandbox | ✅ | ❌ |
This repository supports both macOS and Windows — the two platform builds live side-by-side in the same repo:
- macOS —
Auto_Use/macOS_use - Windows —
Auto_Use/windows_use
Ashish Yadav — founder of Autouse AI
Licensed under the Apache License 2.0 — see LICENSE and NOTICE.
If you use, fork, reference, or derive from this project, you must:
- Preserve the copyright notice and the
NOTICEfile. - Credit Ashish Yadav (Autouse AI) as the original author.
- Link back to the project: https://github.com/auto-use
Yadav, Ashish. Autouse AI — Computer Use. Autouse AI, 2026. https://github.com/auto-use
