Get Suzent running in a few minutes.

1. Install

Use the installer

Download the installer for your platform from the landing page. On macOS, choose Apple Silicon for M-series Macs or Intel for Intel-based Macs. Windows and Linux downloads target x86_64.

On Windows, open the downloaded .exe. On macOS or Linux, open Terminal in your download folder and run the following, replacing <filename> with the downloaded installer's exact name:

chmod +x ./<filename>
./<filename>

Follow the installer to choose your workspace and set up Suzent. Internet access is required. The current release requires Git on macOS and Linux; install it first if the installer asks for it.

Use the terminal

Prerequisites: Node.js 20+ and Git.

powershell -NoProfile -ExecutionPolicy Bypass -Command "irm https://raw.githubusercontent.com/cyzus/suzent/main/scripts/setup.ps1 | iex"

The script installs Suzent and any missing dependencies (Python/uv, Rust, build tools).

Mainland China Mirror Mode

If package downloads are slow from mainland China, enable mirror mode before running setup:

$env:SUZENT_CHINA_MIRROR="1"; powershell -NoProfile -ExecutionPolicy Bypass -Command "irm https://raw.githubusercontent.com/cyzus/suzent/main/scripts/setup.ps1 | iex"

Mirror mode configures PyPI, npm, Playwright, Node via nvm, and Rustup mirrors. If GitHub itself is slow, set SUZENT_REPO_URL or SUZENT_RELEASE_BASE_URL to a mirror you trust before running setup.

2. Launch

suzent start

This starts the backend and opens the desktop interface.

3. Add Your Model Provider

Once the UI is open, go to Settings → Providers to configure your API key. The most common starting points:

Get your key: platform.openai.com/api-keys → Create new secret key (starts with sk-...)

In Settings, open the OpenAI card → API KEYS tab → CHANGE → paste your key. Then go to the MODELS tab → FETCH, select the models you want, and click Save Changes.

Using DeepSeek, Grok, OpenRouter, Ollama, your ChatGPT subscription, or another provider? See Models & Providers.

4. Start Chatting

Pick a model from the model selector in the chat window and send your first message.

That's it. Your agent has memory, tools, and automation ready out of the box. Try asking it to remember something about you, or to research a topic and write a short summary to a file.

Troubleshooting

"Command not found: suzent" — Restart your terminal after installation. If it still fails, check the setup script output for how to manually add the scripts folder to your PATH.

"System Health Check Failed"

suzent doctor

Port conflict on startup — suzent start stops if the port is already in use and tells you which process holds it. Usually that's another copy of Suzent: run suzent stop, or start on another port with suzent start --port <port>.

Updating — Run suzent update, or see Updating Suzent.

More answers are in FAQ & troubleshooting.

Next Steps

Working with your agent: hand it tasks, steer it, and undo what went wrong
Making it yours: what to set up next
Desktop, phone, and browser: reach your agent from your phone or another computer
Models & Providers: add more models and choose which one does what
Memory: what your agent remembers and how to edit it
Tools: everything your agent can do
Chat apps: talk to it from Telegram, Slack, and more
Automation: schedule tasks and check-ins

Quickstart