> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs-parschat.parstechai.com/user-manual/knowledge/manage-sources/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs-parschat.parstechai.com/_mcp/server.

# Manage sources and training status

> See every source your robot learns from, check its training status, retrain it or remove it.

## The Data Collection table

Open your robot and go to **Data Collection** in the training menu. Every source you have added,
whether a file, link or store, is listed in one table:

![Data Collection](/_fern-img/9e0499f964bcc6339adeba668cf53410906737bb9e39f7abaca615002f224b14.webp)

| Column               | What it shows                               |
| -------------------- | ------------------------------------------- |
| **Files**            | The file name or the link address           |
| **Character number** | How much text was extracted from the source |
| **Added on**         | When the source was added                   |
| **Type**             | File, link or store                         |
| **Status**           | The source’s training status                |

The page also shows how many characters you have used out of your plan’s limit.

The **Character number** column is the quickest health check. A source with zero or very few
characters was probably not extracted correctly, for example a scanned PDF with no text layer,
even if adding it looked successful.

## Training statuses

| Status                | Meaning                                                                                                                                                          |
| --------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Learning**          | Training is in progress.                                                                                                                                         |
| **Trained**           | The source was processed and the robot can answer from it.                                                                                                       |
| **Training error**    | Processing failed. The robot cannot use anything in this source.                                                                                                 |
| **Out of Characters** | Your plan’s character allowance is used up, so the source was not trained. [Upgrade your plan](/user-manual/account/plans) or remove sources you no longer need. |

The counters at the top of the page add up how many sources are in each status.

> **Note**
>
> **Statuses do not update on their own.** Refresh the training status to see the latest state.

If your robot does not answer a question it should know, check the statuses here first. A source
in **Training error** is invisible to the robot.

## Retrain a source

When a file or web page has changed, or a previous training attempt failed, you don’t need to delete and
re-add it. Select the sources and click **Retrain**. When it finishes, the panel shows
*"Retrain success"*.

Check the statuses again afterwards and confirm every source is **Trained**.

## Delete a source

To remove a source you no longer need, click **Delete** in its row. Its content leaves the robot’s
knowledge, and the robot stops answering from it.

> See every source your robot learns from, check its training status, retrain it or remove it.