> For the complete documentation index, see [llms.txt](https://docs.praiz.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.praiz.io/query-and-analyze/insights.md).

# Insights

Analyze all of your conversations from the data extracted by one agent.

Insights analyze the data one agent has extracted across all of your calls. You choose the agent, the period and the context of the analysis. Praiz generates a report from everything that agent has extracted. For example: the most frequent objections, the most cited competitors, or any other analysis based on the agent's fields.

{% hint style="info" %}
Before you start: an agent applied to at least 3 calls. Insights work on the data that agent has extracted.
{% endhint %}

## Create a report

Open "Insights" in the left menu. Each report sits in its own tab. Click "+" to create a tab, or open an existing one.

![Insights page with report tabs and settings](https://3785996597-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FU54oKzNb3EyinriyUP5P%2Fuploads%2FQcvx8JEZrOGSQ5mdPJIK%2F26-insights-top.png?alt=media)

{% stepper %}
{% step %}

### Agent

Under "Select an agent", choose the agent whose data you want to analyze.

{% hint style="info" %}
Generation needs at least 3 rows of data for that agent ("Generate Insights requires at least 3 rows in Agent data").
{% endhint %}
{% endstep %}

{% step %}

### Period

Select the range to cover: week, month, quarter, or a custom range.
{% endstep %}

{% step %}

### Filters (optional)

Narrow the analysis to part of your calls: people, teams, the Aircall line used to call, or the value of a field the agent extracted.

The number of calls that match your filters appears as soon as you change a filter, before you run the report. Your filters are kept when you run the report again.
{% endstep %}

{% step %}

### Configure

Click "Configure Insights" to set the report options: language, length, whether to include charts and emojis, and free context to steer the analysis ("Context for AI"). The context you type is kept after generation: you can run the report again without retyping it.

![Configure Insights window with the report settings](https://3785996597-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FU54oKzNb3EyinriyUP5P%2Fuploads%2Fke71Ly6SYoFdUW6OsUL4%2F27-insights-settings.png?alt=media)
{% endstep %}

{% step %}

### Generate

Click "Generate Insights".
{% endstep %}
{% endstepper %}

What to expect: the report appears in the tab. You can read it, copy it ("Copy") or download it as a PDF ("Download as PDF"). When the report cites a call, the link opens that recording. Tabs are kept, so you keep several analyses side by side, and running a second report does not interrupt the first one.

## Export the agent's raw data

Below the report, the page lists everything the agent extracted over the period, call by call (30 records per page, "Previous" and "Next" buttons). Click "Download CSV" at the bottom of the page to download all of that raw data as a CSV file.

![Download CSV button below the agent data list](https://3785996597-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FU54oKzNb3EyinriyUP5P%2Fuploads%2FOifDqymQOXRaUU9IyFJi%2F29-insights-download-csv.png?alt=media)

## Example analyses

| Question                                  | What Insights show you                                         |
| ----------------------------------------- | -------------------------------------------------------------- |
| Which objections come up most?            | The most frequent objections, and how they trend               |
| Which competitor costs us the most deals? | Competitive pressure measured across all of your conversations |
| What are our customers asking for?        | Product feedback consolidated by theme and by frequency        |
| Where are our customer risks?             | Churn signals detected across the book of business             |

## Receive the report automatically

Click "Schedule recurring insights" to schedule the report. Praiz generates it and sends it by email or on Slack, at the frequency you choose. See "Recurring Insights".

## Insights or Analytics

Insights are about the content of your conversations. The "Analytics" page is about how Praiz is used: call duration and count, video versus phone split, active users, monthly trend and usage per user, with a CSV export. See "Analytics".

## Going further

To query your conversational database from Claude, ChatGPT or Gemini, see "MCP server".