> 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/run-an-agent.md).

# Run an agent

## Overview

Once your agent is created (and ideally [backtested](/agents/backtest-an-agent.md)), you can **launch it in execution**. An agent always runs against a [**portfolio**](/portfolios/portfolios.md), in one of two modes:

* **Demo (paper trading)** — the agent runs on a **demo portfolio** with virtual capital. Ideal to observe its behavior in real time without risk.
* **Live** — the agent runs on a **real portfolio** connected to a broker, placing real orders.

{% hint style="info" %}
**Live** execution requires a **Pro or Max** plan and depends on your region and broker. See [Availability by region](/platforms/availability-by-region.md).
{% endhint %}

***

## Launching an execution

From an agent's page, click **Activate agent**. A dialog guides you through the necessary steps.

### 1. Portfolio

Choose the **portfolio** the agent will run on:

* A **demo portfolio** — virtual capital, for paper trading.
* A **real portfolio** — a connected broker account, for live trading *(requires a Pro or Max plan and an eligible region)*.

You can create a new demo or real portfolio directly from this step. See [Portfolios](/portfolios/portfolios.md). The portfolio's starting capital (demo) or connected broker (real) is used automatically — there's no separate capital or API-key step.

{% hint style="info" %}
Agents that only **monitor or alert** (no orders) don't need a portfolio. Obside resolves the assets to watch and skips straight to notifications.
{% endhint %}

### 2. Exchange *(live, multi-venue brokers only)*

If your connected broker exposes several venues, pick the one these assets should trade on.

### 3. Notifications *(if the Notifications capability is enabled)*

Select at least one **channel** to receive notifications from the agent (**Telegram, push or email** — WhatsApp coming soon). See [Notification channels](/alerts-and-notifications/channels.md).

### 4. Summary

Review the mode, portfolio and settings, then **start** the agent. Live trading agents ask you to confirm a short risk disclaimer before activation.

***

## Monitoring the execution

Once the agent is running, you can access several tabs to track its activity in real time:

* **Metrics** — cumulative performance since the start.
* **Performance** — value/PnL curve since activation.
* **Chart** — TradingView visualization of entries and exits.
* **Trades** — list of open and closed positions.
* **Activity** — log of the agent's decisions (useful for AI traders and agents with the News capability). *Availability may vary.*

{% hint style="info" %}
The **maximum number of simultaneous executions** depends on your subscription plan. See [Pricing](/misc/pricing.md).
{% endhint %}

***

## Start, pause, stop

You can **pause** or **stop** an execution at any time from the agent's page. A stopped execution preserves its viewable history, but no longer generates new trades or notifications.

{% hint style="success" %}
**Best practice**: before launching an agent in live mode, run it in demo mode for a few days or weeks to validate its behavior in current market conditions.
{% endhint %}
