> 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/introduction.md).

# Introduction

## What is an agent?

An **agent** is an automation you create by describing your intent in natural language. On Obside, anything that runs on your behalf — a scheduled trade, a trading bot, an AI trader, a smart alert — is an agent.

Each agent has:

* A **description** written in natural language (your intent).
* A **code** automatically generated by Obside's AI, formalizing this description.
* One or more **assets** it operates on.
* One or more **capabilities** enabled (trading, notifications, AI, news).
* A history of **backtests** and **executions**.

***

## Types of agents

Depending on what you describe in the chat, the agent takes a different form:

| Type                     | Description                              | Typical capabilities |
| ------------------------ | ---------------------------------------- | -------------------- |
| **Scheduled trade**      | A one-off operation to execute           | Trading              |
| **Trading bot**          | A strategy following deterministic rules | Trading              |
| **AI trader**            | An autonomous agent powered by an LLM    | Trading + AI + News  |
| **Smart alert**          | A condition that notifies you            | Notifications        |
| **Portfolio monitoring** | Continuous tracking of your positions    | Notifications        |

{% hint style="success" %}
You can **combine multiple capabilities** within a single agent — for example, a trading bot that also notifies you of every trade.
{% endhint %}

***

## The lifecycle of an agent

1. **Creation** — you describe your intent in [Obside Chat](/obside-chat/introduction.md). The AI rephrases, you confirm, and the agent is created.
2. **Backtest** *(for trading agents)* — you test the agent on historical data to evaluate past performance. See [Backtest an agent](/agents/backtest-an-agent.md).
3. **Execution** — you run the agent on a [**portfolio**](/portfolios/portfolios.md), in **demo** (paper trading) or **live** mode. See [Run an agent](/agents/run-an-agent.md).
4. **Monitoring & adjustments** — you review performance, trades, triggered notifications, and iterate via the chat linked to the agent.

{% hint style="info" %}
All your agents are accessible from the **Agents** section in the sidebar. You can organize them in folders, filter by capability or status, and find them via search.
{% endhint %}

To go further:

* [Agent capabilities](/agents/capabilities.md)
* [Create an agent](/agents/create-an-agent.md)
* [Backtest an agent](/agents/backtest-an-agent.md)
* [Run an agent](/agents/run-an-agent.md)