Activated Cloud

Documentation

First boot

Plug it in, answer six questions, and your agent is running. No terminal, no config files.

The first time the device starts, it opens a setup wizard instead of the dashboard. You cannot skip past it — the agent needs a name, a brain, and a lock before it is any use — but it takes about two minutes. Every screen below is the real thing.

1. Welcome

The language selector sits in the top right. Change it here and the whole wizard — and the dashboard afterwards — switches immediately. Seventeen languages ship with the device.

Set your language before you start and everything that follows is translated.

2. Name your agent

Two things: what you want to call your agent, and what it should call you. It answers to the name you give it and greets you by yours — this is not decoration, it is written into the agent's identity.

Both names can be changed later, and you can simply tell your agent to go by something else.

3. Give it a brain

Pick a model provider and paste an API key. Nothing is pre-selected — the choice is yours, and the key is stored on the device. Selecting a provider opens a panel where the key and the model go.

You can skip this and connect a model later from Models — but the agent cannot answer anything until you do.

4. Choose what it can do

A grid of capabilities: running commands, browsing, reading and writing files, remembering, and so on. Everything starts switched off. Turn on what you want — at least one is required — and a selected tile turns black so the set you have chosen is obvious at a glance.

This is a real boundary, not a preference. An agent without the terminal capability cannot run commands, whatever it is asked.

5. Decide what needs your OK

How the agent behaves when it wants to change something. Ask every time is the most cautious; the middle option lets it read freely but check before anything destructive; the last lets it act on its own.

Changeable at any time in Config. Start cautious — you can loosen it once you trust the setup.

6. Let it remember

Whether the agent keeps what it learns between conversations. With memory on it gets more useful the longer you use it — preferences, facts about you, the way you like things done. With it off, every conversation starts from nothing.

See Memory for how to steer what it keeps and how to make it forget something.

7. Reach it from anywhere

Optionally connect a messaging channel so you can talk to your agent from your phone. This is the one step you can skip outright — channels are easy to add later from the dashboard.

Skip it now if you like — Channels covers the full set and how to connect each one.

8. Make it yours

Your name, and the password that locks the screen. The agent uses your name to address you; the password is what unlocks the device when the screen is locked.

Both password fields must match before Finish setup becomes available.

Keep the password

It unlocks the screen and there is no reset link — this is your device, not an account on ours. Store it somewhere you trust.

9. Done

A summary of everything you chose, so you can check it before you start. Your agent is named, has a brain, knows what it may do, and is locked.

From here you land in Chat — see Your first conversation for what to try first.

After setup

You land in Chat, which is where you will spend most of your time. The dock along the bottom takes you everywhere else — Sessions, Skills, Models, Channels — and the grid icon opens the full list of destinations.

Chat is home. Type at the bottom, press / for commands, or the microphone to dictate.

Light or dark

The same screen in the other theme. Switch between them from the dock at any time — the whole system follows, including the setup wizard and the lock screen.

Every page ships in both themes; nothing is light-only.

Nothing here is permanent. Every answer can be changed afterwards from the dashboard, and adding a capability later is one click.