> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rerun.build/llms.txt
> Use this file to discover all available pages before exploring further.

# Quickstart

> From signing up to a working agent.

## Before you start

You need an account on [app.rerun.build](https://app.rerun.build) and a card on file. Rerun starts with a 7 day free trial and you can cancel anytime.

Nothing to install. Everything happens in the browser.

## 1. Sign up

Open [app.rerun.build](https://app.rerun.build). You land on the board, the isometric view where your Boxes and agents live, with a sign-in dialog on top.

Right after signing up you answer three short questions about your sector, your team size and your role. They only decide which templates you get shown first.

## 2. Pick your first agent

The template library opens on its own, filtered on the sector you picked. Two ways forward:

<Tabs>
  <Tab title="Install a template">
    A template is a setup that already works: a prompt, skills, scheduled tasks and the connectors it needs. Pick one, and Rerun creates the agents it contains in your account.
  </Tab>

  <Tab title="Start from scratch">
    Click **Start from scratch**, give the agent a name, and describe what you want it to do in the chat. It writes its own prompt from your description.
  </Tab>
</Tabs>

## 3. Place it on the board

The toolbar switches to placement mode. Tap a tile and the agent takes its spot.

Your first agent needs a Box, and Rerun claims one for you from a pool of ready machines. It takes a few seconds. A Box is a private cloud computer, isolated from every other account, and it will host every agent you place in it.

## 4. Run the guided setup

An agent that came from a template usually needs to be connected to your own accounts. When it does, the chat opens on a **Start installation** button instead of a message box.

Click it and the agent walks you through its own setup: it tells you what it needs, offers to install the connectors it is missing, and asks you for the values only you can give. You answer in the chat, exactly like a conversation.

<Note>
  When it needs an API key, the agent opens a masked form. The value goes straight to your Box and is never shown to the agent or stored in the conversation. Never paste a secret into the chat itself.
</Note>

When the setup is done, the agent switches on the scheduled tasks and webhooks it shipped with, and tells you what it can do and when it will next run.

## 5. Talk to it

You now have a normal message box. Ask it for something.

While it works you see every step: the files it reads, the searches it runs, the tools it calls. If it needs a decision, it stops and asks you, right in the chat.

## 6. Let it run without you

Two ways to make an agent work while you are away.

<CardGroup cols={2}>
  <Card title="Scheduled tasks" icon="clock" href="/agents/automations#scheduled-tasks">
    A cron and an instruction. Every Monday at 9, summarise last week and email it to me.
  </Card>

  <Card title="Webhook triggers" icon="webhook" href="/agents/automations#webhook-triggers">
    A public URL. When Stripe posts a refund, handle it.
  </Card>
</CardGroup>

You can add both from the agent panel, or just ask the agent to set them up for you.

## Next steps

<CardGroup cols={3}>
  <Card title="Core concepts" icon="book-open" href="/concepts" horizontal />

  <Card title="Connect your tools" icon="link" href="/agents/connectors" horizontal />

  <Card title="Drive it from the API" icon="code" href="/api/introduction" horizontal />
</CardGroup>
