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

# Quickstart

> Short builds that get you from an empty account to an agent taking real calls, then better ones.

Four guides, each one a working thing at the end rather than a tour.

Start with whichever matches what you are trying to do. They do not have to be done in
order, though the campaign one assumes you already have an outbound agent.

***

## Pick a starting point

<Steps>
  <Step title="Someone calls you">
    **[Create your first inbound agent](./first-inbound-agent)**

    An agent that answers your number, works out why the caller rang, takes their
    details, and transfers to a person when asked.
  </Step>

  <Step title="You call someone">
    **[Create your first outbound agent](./first-outbound-agent)**

    An agent that calls a lead, checks interest, and books either a meeting or a
    callback.
  </Step>

  <Step title="You call a list of people">
    **[Set up your first campaign](./first-campaign)**

    A deliberately tiny campaign, two contacts, so you see every moving part before
    pointing one at real numbers.
  </Step>

  <Step title="It works, but it is not good yet">
    **[Improving an agent](./improving-an-agent)**

    The loop that turns a working agent into a reliable one. Break it, capture the
    failure as a test, fix it by talking to it, then prove the fix held.
  </Step>
</Steps>

***

## Before any of them

You need an account. [Sign up here](https://platform.fluents.ai/signup).

New accounts start on a **trial**, which you can see under **Billing** along with what
each paid plan includes. Check there for your current allowance rather than trusting a
number written in a doc.

<Tip>
  Do the inbound guide first even if your real goal is outbound. It is shorter, you can
  test it by phoning your own number, and everything it teaches about prompts and
  voices carries straight over.
</Tip>

***

## What to read after

Once something works, these are the pages worth your time.

* **[Agents](/product/features/agent)** for every setting the quickstarts skipped.
* **[Testing an agent](/product/features/agent-testing)** for turning a bad call into
  a fix you can prove.
* **[Calls](/product/features/call)** for reading what actually happened.
* **[Post-call](/product/features/post-call)** for pulling structured results out of
  conversations.