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

# Visualizations

> Build charts and dashboards from your collections

<Frame>
  <img src="https://mintcdn.com/prismeai/9aPAXX2QoCsMDyNB/images/ai-collection-visualize.png?fit=max&auto=format&n=9aPAXX2QoCsMDyNB&q=85&s=c63563bf9198f8e881a279bf63e5b679" alt="Collection visualizations" width="1516" height="826" data-path="images/ai-collection-visualize.png" />
</Frame>

Visualizations turn your collections into charts — distributions, trends, and comparisons — without exporting anything to a spreadsheet. Charts live in two places:

* The **Visualize** tab of a collection — charts built from that collection's data
* The **Visualizations** page — all your saved charts across collections, in one dashboard

## Chart Types

| Type        | Best for                                | Example                         |
| ----------- | --------------------------------------- | ------------------------------- |
| **Bar**     | Comparing categories                    | Deal value by pipeline stage    |
| **Line**    | Trends over time                        | NPS score by month              |
| **Pie**     | Distribution of a whole                 | Sentiment breakdown of feedback |
| **Scatter** | Correlation between two numeric columns | Price vs. stock level           |
| **Heatmap** | Density across two dimensions           | Errors by service and severity  |

## Creating a Chart

<Steps>
  <Step title="Open the Visualize tab">
    From a collection's detail page, switch to **Visualize** and click **New Chart**.
  </Step>

  <Step title="Choose a chart type">
    Pick Bar, Line, Pie, Scatter, or Heatmap.
  </Step>

  <Step title="Map your columns">
    * **X Axis** — the column to spread across (a category or a date)
    * **Y Axis** — the numeric column to measure
    * **Group By** (optional) — split the measure into series by another column
  </Step>

  <Step title="Pick an aggregate">
    Choose how values are combined: **Sum**, **Average**, **Count**, **Min**, or **Max**.
  </Step>

  <Step title="Name and create">
    Give the chart a title and click **Create**. It appears in the collection's Visualize tab and on the global Visualizations page.
  </Step>
</Steps>

<Tip>
  Charts compute over the **entire collection**, using the same exact aggregation engine agents use. A "Sum of deal\_value by stage" bar chart and an agent answering "total pipeline by stage" produce the same numbers.
</Tip>

## Example Charts

A few combinations that work well as a starting dashboard:

* **Sales pipeline** — Bar chart: sum of `deal_value` by `stage`; Bar chart: count of deals by `owner`
* **Customer feedback** — Pie chart: count by `sentiment`; Line chart: average `nps_score` by month
* **Product catalog** — Bar chart: count of products by `category`
* **Error logs** — Line chart: error `count` by day; Heatmap: errors by `service` and `severity`

## Best Practices

<AccordionGroup>
  <Accordion title="Aggregate on clean categories">
    Charts group by exact values. Normalize category columns (consistent casing and wording) before building charts, or you will see duplicate bars for the same logical category.
  </Accordion>

  <Accordion title="Prefer Count when values are missing">
    Sum and Average silently skip empty cells. If a numeric column has many gaps, a Count chart gives a more honest picture of your data.
  </Accordion>

  <Accordion title="Keep dashboards focused">
    Three to six charts per topic is usually enough. If a chart never changes your decisions, remove it.
  </Accordion>
</AccordionGroup>

## Next Steps

<CardGroup cols="2">
  <Card title="Working with Data" icon="table" href="/products/ai-collection/data-management">
    Clean your data before charting it
  </Card>

  <Card title="Using Collections with Agents" icon="robot" href="/products/ai-collection/agents">
    Let agents run the same aggregations conversationally
  </Card>
</CardGroup>