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

# Triggers in Agents

> Start Agent runs automatically on a schedule, an app event, a change on the web, or a change in your Profound data

A run trigger starts an Agent run automatically when its condition is met. The condition can be a schedule, an event in a connected app or on the web, or a change in your Profound data.

Add triggers in the Agent's settings panel, or ask Aim in the Agent's chat to add them.

## Trigger types

| Trigger | Runs when | Example |
| - | - | - |
| **Schedule** | A scheduled time arrives | Every Monday at 9:00 AM |
| **App** (listed by app name, for example **Slack**) | An event happens in a connected app | Someone posts a new message in a Slack channel |
| **Monitor** | Something changes on the web | A competitor publishes a new blog post |
| **Profound data** | A Profound metric meets a condition | Visibility Score drops 10% compared with the last 7 days |

Add more than one trigger to an Agent, and it runs when any trigger's condition is met. By default, an Agent has no triggers and runs only when you select **Run Agent**.

## How triggers behave

* **Each trigger has its own switch.** Pause or resume a trigger with its switch under **Triggers** in the settings panel. Only the person who created a trigger can switch it.
* **Triggered runs use the current version of the Agent.** Changes you make apply to the next triggered run.
* **Triggers belong to the person who set them.** Only that person sees the triggered run and its outputs. People the Agent is shared with can run it themselves and see their own runs. Triggers aren't saved in version history.
* **Every triggered run is a chat.** Find it under **Past Runs** at the top of the Agent's page, with any outputs under **Artifacts**.
* **Triggered runs can run Workflows, too.** If the Agent's instructions include a Workflow, a triggered run runs it and uses its results, the same as a manual run. The person who set the trigger needs access to the Workflow.

## Set default values for inputs

A triggered run starts without anyone there to fill in the Agent's inputs. Set a value for each required input on the trigger so the Agent can run on its own. When creating a Schedule, Monitor, or Profound data trigger that needs input values, select **Next**, fill in **Input values**, then select **Add trigger**. Saved dialogs for these triggers have **Configuration** and **Default inputs** tabs.

If a required input has no default value, Profound shows a warning for the trigger. Aim can also set up default values for you: ask it in the Agent's chat.

## Add a schedule

In the settings panel, select **Add** next to **Triggers** and point to **Schedule**. Select **Hourly**, **Daily**, or **Weekly** to add a schedule that starts now. If inputs need values, the preset opens the form first. Select **Custom** to set up a schedule yourself:

1. Set **Starts on** to the first date and time the Agent should run.
2. Select a **Frequency**: **Doesn't repeat**, **Daily**, **Weekdays**, **Weekend**, **Weekly**, **Monthly**, **Annually**, or **Custom**. For a custom cadence, select **Custom**, set **Repeat every** (for example, 1 week or 4 hours), and select the days under **Repeat on**.
3. For a custom hourly or minute schedule, optionally enable **Only run between** under **Active hours** and set **From** and **Until**.
4. Select **Add trigger** to save. If inputs need values, select **Next**, fill in **Input values**, and then select **Add trigger**.

The trigger appears under **Triggers**, named after how often it runs, for example, "Every Monday at 9:00 AM". Use its switch to pause it. Select the trigger to edit it or select **Remove** in its dialog to delete it.

To add a schedule in the chat, ask Aim, for example, "Run this Agent every Monday at 9 AM."

## Add an app trigger

1. Select **Add** next to **Triggers** and point to an app listed in the menu, for example **Slack**.
2. Select the event, for example **New channel message**, **Reaction added to a message**, or **User mentioned**.
3. Complete the trigger sentence. For Slack messages, select which messages count and the sender (**Anyone** or **A specific person**). For reactions, select the reaction and the person who reacted. With **User mentioned**, select the person who gets mentioned, then select who mentions them (**Anyone** or **A specific person**). Select **including replies** or **excluding replies** for messages and mentions. Keep **All channels** (the default for these events) or select a channel. To watch all channels again, select the X next to the channel. With **All channels**, you must select a specific person who sends the message, adds the reaction, or gets mentioned.

The trigger saves as soon as the inline sentence is complete. If the Agent needs input values, a dialog opens instead: add the event, complete its sentence and **Input values**, then select **Save**.

Slack triggers can start a run when a specific person sends a message, reacts to a message, or is @mentioned. For example, add a trigger that fires when you're @mentioned in a channel, and tell the Agent to reply for you. For the Agent to reply in Slack, add Slack as a tool to the Agent.

In the channel list, search to find any of your channels. Private channels are hidden until you message @Profound in Slack and link your account. Learn more in [Connect Slack to Profound](https://docs.tryprofound.com/integrations/slack/connect-slack-to-profound.md).

## Add a monitor

A monitor watches the web for updates that match a query you write. Profound checks for new matches on the schedule you choose, ignores anything it has already reported, and starts a run only when something new appears.

1. Select **Add** next to **Triggers** and select **Monitor**.
2. Optionally, enter a **Name** (up to 200 characters), for example "Competitor blog monitor". If you leave it empty, Profound names the monitor "Monitor 1", "Monitor 2", and so on.
3. In the **Query** field, describe what to watch (up to 2,000 characters), for example, "a new post on competitor.com/blog" or "changes to the pricing page at competitor.com/pricing".
4. Select a **Frequency**: **Every hour** (the default), **Every 2 hours**, **Every 4 hours**, **Every 6 hours**, **Every 12 hours**, or **Every day**.
5. Select **Add trigger** to save. If inputs need values, select **Next**, fill in **Input values**, and then select **Add trigger**.

After you save a monitor, you can rename it, but you can't change the query or frequency. To change them, remove the monitor and add a new one.

### Writing a good monitor query

* Describe what you're looking for in plain language, the way you'd brief a colleague.
* Keep your query clear and specific: include what to watch for and where to watch for it.
* Leave out date ranges. The monitor already surfaces only what's new.

Prefer the queries in the first column and avoid the ones in the second.

| Prefer ✅ | Avoid ❌ |
| - | - |
| AI startup funding announcements | Find all AI funding news from the last 2 years |
| Tesla news and announcements | Tesla news after December 12, 2025 |
| New blog posts published on competitor.com | competitor.com blog |

## Add a Profound data trigger

1. Select **Add** next to **Triggers** and select **Profound data**.
2. Select the **Category** and the **Asset** to watch. FactCheck metrics have no **Asset** field.
3. Build the **Trigger** condition: set **When** a metric **drops by**, **rises by**, **drops below**, or **rises above** a value (in % or a number), **Against** a comparison window (**Last day**, **Last 7 days**, **Last 14 days**, **Last month**, or **Last 3 months**).
4. Narrow the scope with **Filters**: Topics, Platforms, Regions, Tags, or Personas.
5. Set the **Check frequency** (**Daily** by default, or **Weekly**, **Bi-weekly**, or **Monthly**) and select **Add trigger**. If inputs need values, select **Next**, fill in **Input values**, and then select **Add trigger**.

Available metrics:

* **Visibility**: Visibility Score, Share of Voice, Average Position, and their ranks.
* **Sentiment**: positive percentage, negative percentage, and assessment count.
* **Citations**: Citation Share, citation count, Citation Rank, and Citation Coverage.
* **FactCheck**: Accuracy score, and inaccurate, accurate, and verified claim counts. These appear only when FactCheck is set up for the category.

## Edit, pause, or remove a trigger

Under **Triggers** in the settings panel:

* **App triggers:** edit the inline sentence. If the trigger opens in the app dialog, use **Save** to commit additions or removals. Saved sentences in that dialog are read-only; remove and recreate them to change the condition.
* **Schedule and Profound data:** select the trigger, change its settings, and select **Save**, or select **Remove** to delete it.
* **Monitor:** select the trigger to rename it or change its **Default inputs**, then select **Save**. To change its query or frequency, remove it and add a new monitor.

To pause a trigger without deleting it, turn off its own switch. A disabled trigger shows **Disabled** when you hover over it. Turn the switch back on to resume. To stop an Agent from running on its own permanently, remove its triggers.

Trigger changes apply right away.
