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

# Schemas

> Define the structure of your extracted data with column definitions.

A schema defines the columns that Lasso extracts from your source data. Think of it as a blueprint for your table.

## Column definitions

Each column has a key, label, and type:

```json theme={null}
{
  "key": "product_name",
  "label": "Product Name",
  "type": "text",
  "required": true
}
```

* **key** -- Machine-readable identifier (snake\_case). Used to access data in rows.
* **label** -- Human-readable name shown in exports and the dashboard.
* **type** -- The data type Lasso should extract. See below.
* **required** -- Whether this column must have a value for every row.

## Column types

| Type       | Description              | Example value                              |
| ---------- | ------------------------ | ------------------------------------------ |
| `text`     | Plain text               | `"iPhone 15 Pro"`                          |
| `number`   | Integer or decimal       | `999.99`                                   |
| `url`      | Valid URL                | `"https://example.com"`                    |
| `email`    | Email address            | `"info@example.com"`                       |
| `date`     | Date or datetime         | `"2025-03-10"`                             |
| `boolean`  | True or false            | `true`                                     |
| `richtext` | HTML or markdown         | `"<p>Description</p>"`                     |
| `enum`     | One of predefined values | `"electronics"`                            |
| `tags`     | Array of strings         | `["wireless", "bluetooth"]`                |
| `image`    | Single image URL         | `"https://cdn.example.com/img.jpg"`        |
| `images`   | Array of image URLs      | `["https://...1.jpg", "https://...2.jpg"]` |
| `json`     | Arbitrary JSON           | `{"specs": {"weight": "187g"}}`            |

## Auto-generated schemas

If you are not sure what columns to define, you can let the AI generate a schema from sample data:

<CodeGroup>
  ```typescript TypeScript theme={null}
  const schema = await client.schemas.generate({
    sample_data: "Product: iPhone 15 Pro, Price: $999, Color: Titanium",
    name: "Smartphones",
  });

  console.log(schema.columns);
  ```

  ```python Python theme={null}
  schema = client.schemas.generate(
      sample_data="Product: iPhone 15 Pro, Price: $999, Color: Titanium",
      name="Smartphones",
  )

  print(schema["columns"])
  ```
</CodeGroup>

The AI analyzes your sample and infers appropriate column keys, labels, and types.

## Reusing schemas

Schemas are reusable across multiple tables. Create a schema once, then reference it by ID when creating new tables. You can update a schema at any time -- existing tables retain their original column definitions.