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

# Alerts & Reports

> Set up threshold-based alerts to stay notified about your agent's activity, and generate reports for stakeholder sharing.

**Alerts & Reports** is where you configure proactive notifications and generate performance reports for your agent. Use this section to stay on top of activity spikes, performance dips, and other conditions that require your attention.

<Note>
  Alerts & Reports is a **Beta** feature.
</Note>

## Alerts

The Alerts section has two tabs:

### Configuration

This is where you create and manage alert rules. An alert configuration defines:

* **Title** — a descriptive name for the alert (e.g. `High escalation rate alert`)
* **Alert type** — the metric or event being monitored
* **Threshold** — the numeric value that triggers the alert (must be 0 or greater)
* **Unit** — the measurement unit for the threshold (e.g. conversations, percentage, minutes)

When the defined condition is met, the alert fires and is recorded in the **History** tab.

#### Creating an alert

<Steps>
  <Step title="Go to Alerts & Reports">
    Navigate to **AI Agent > Evaluate > Alerts & Reports** in the left sidebar.
  </Step>

  <Step title="Open the Alerts tab">
    The page opens on the Alerts section by default. Click **New Alert** in the Configuration sub-tab.
  </Step>

  <Step title="Define the alert">
    Fill in the title, select the alert type, set your threshold value, and choose the unit.
  </Step>

  <Step title="Save">
    Click **Save** to activate the alert. It will start monitoring immediately.
  </Step>
</Steps>

### History

The History tab shows a log of all alerts that have been triggered, including:

* Which alert fired
* When it was triggered
* The value that crossed the threshold

Use the History tab to audit alert activity and understand how often specific conditions are occurring.

## Reports

The Reports section lets you generate summaries of your agent's performance over a defined period. Reports are useful for:

* Sharing performance snapshots with stakeholders or leadership
* Monthly or quarterly reviews
* Tracking trends over time

### Report configuration

When creating or editing a report configuration, you can define:

* **Name** — a label for the report
* **Enabled** — toggle the report on or off
* **Language** — choose the language for the generated report. Supported languages are **English, Spanish, and Portuguese**.
* **Recipients** — email addresses that will receive the report
* **Modules** — the sections to include in the report

Click **Generate Report** to create a new report. You can select the date range and the metrics to include.

### Available modules

When configuring which sections to include in a report, you can choose from the following modules:

| Module                    | Description                                                        |
| ------------------------- | ------------------------------------------------------------------ |
| **Home**                  | General overview and key metrics for your agent                    |
| **Hotspots**              | Top topics and issues ranked by priority score                     |
| **Recommendations**       | AI-generated suggestions to improve your agent                     |
| **Escalations**           | Data on conversations that were escalated to human agents          |
| **Live Chat Performance** | Performance metrics for your live chat / contact center operations |
| **Billing**               | Usage and billing summary                                          |

<Tip>
  Include the **Escalations** and **Live Chat Performance** modules in your weekly report to give your contact center team a complete picture of how AI-to-human handoffs are performing.
</Tip>

<Tip>
  Set up a **high escalation rate alert** as your first alert. Escalations are often the clearest signal that the agent is struggling with a specific topic—and catching them early lets you fix the issue before it impacts more users.
</Tip>