> For the complete documentation index, see [llms.txt](https://docs.obside.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.obside.com/agents/create-an-agent.md).

# Create an agent

## Creation through natural language

On Obside, you create an agent by **describing your intent in the chat**. No more step-by-step forms: you write what you want to achieve, and the AI takes care of turning your description into a ready-to-use agent.

The process unfolds in a few steps:

1. **You describe your intent** in Obside Chat — what to automate, on which assets, with which rules.
2. **The AI rephrases** your request to make sure it's correctly understood. You can correct or clarify if needed.
3. **The agent is created**: its description, code, assets and capabilities are defined automatically.
4. **You can iterate** through the chat linked to the agent to refine, test or optimize it.

***

## The building blocks of an agent

To write your request well, it helps to understand what an agent is made of. The following pages detail each element:

* [**Description in natural language**](/agents/create-an-agent/description.md) — how to phrase your intent so it's faithfully interpreted.
* [**Assets**](/agents/create-an-agent/assets.md) — selecting the instruments the agent operates on.
* [**Entry rules**](/agents/create-an-agent/entry-rules.md) — when and how the agent opens a position.
* [**Exit rules**](/agents/create-an-agent/exit-rules.md) — when and how the agent closes a position.
* [**Rules reliability**](/agents/create-an-agent/rules-reliability.md) — how Obside guarantees the agent does what you described.
* [**Example requests**](/agents/create-an-agent/examples.md) — concrete examples for trading bots, investing, alerts and market analysis.
* [**Indicators list**](/agents/create-an-agent/indicators-list.md) — reference of available technical indicators.

{% hint style="success" %}
**Tip**: if you're out of inspiration, the chat offers **idea suggestions** in each of the three families of use cases (automation, alerts, market understanding). Click a suggestion to start from an example and adapt it to your needs.
{% endhint %}