> ## 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.

# Tables & extraction

> Understand how Lasso extracts structured data from your source files.

Tables are the core resource in Lasso. A table represents an extraction job that takes unstructured source data and produces structured rows of product information.

## How extraction works

When you create a table, Lasso processes your source files through an AI pipeline:

1. **Document parsing** -- PDFs, spreadsheets, and images are parsed to extract raw text and visual elements.
2. **Schema mapping** -- The AI maps the raw content to your schema's column definitions.
3. **Row generation** -- Each detected product or item becomes a row with values for each column.
4. **Validation** -- Extracted values are validated against column types (numbers, URLs, emails, etc.).

## Source types

You can provide data to extract from in three ways:

| Source    | Description                                                                                       |
| --------- | ------------------------------------------------------------------------------------------------- |
| **Files** | Upload PDFs, spreadsheets, images, or documents via the [Files API](/api-reference/files/upload). |
| **URLs**  | Pass publicly accessible URLs directly when creating a table.                                     |
| **Text**  | Provide raw text content for extraction.                                                          |

## Table lifecycle

A table moves through these statuses:

```
queued → processing → completed
                    → failed
```

* **queued** -- The job is waiting to be picked up.
* **processing** -- Extraction is actively running. The `progress` field (0-100) tracks completion.
* **completed** -- All rows have been extracted and are ready to query.
* **failed** -- Something went wrong. Check `error_message` for details.

## Polling vs webhooks

You have two options to know when extraction finishes:

* **Polling** -- Use `client.tables.waitForCompletion()` (SDK) or poll `GET /v1/tables/{id}` until status is `completed`.
* **Webhooks** -- Pass a `webhook_url` when creating the table. Lasso sends an HTTP POST when processing completes.

## Working with rows

Once a table is completed, each extracted item is a row. Rows contain:

* **data** -- A key-value object matching your schema columns.
* **validation\_status** -- Whether the row passed type validation.
* **enhancement\_status** -- Per-column status of any AI enhancements.
* **is\_edited** -- Whether the row was manually modified via the API.

Rows can be updated individually, in bulk, or deleted. See [Rows API](/api-reference/rows/list) for details.