The Data Source is the centralised hub of information that empowers your Virtual Assistant to provide precise and valuable information to your visitors. The virtual assistant can deliver accurate responses and offer relevant suggestions by uploading your data.
📘
Reminder of Roles & Permissions
Only an Aministrator or any custom role with permission of Manage Destination is allowed to add, view, edit, download and delete the data source.
To add a data source:

📘
The Data Source Name must be unique.

You can select one of the data source types: CSV, PDF, URL or Markdown File.

Here is the sample CSV file (Table 1) with not less than 20 MB using 1 language; all text is readable and not encrypted, fewer than 20,000 rows, and each column has a title.

In this sample data source:
📘
CSV File Data Source Reminder
- The maximum file size is 20 MB.
- Using UTF-8 as a file format for CSV is recommended.
- CSV files are limited to 20,000 rows. Scrapping will fail if it exceeds this limit.
- One language per data source is recommended.
- Images or multiple text columns won't be scraped.
- Encryption or password-protected files are not supported.
- For accurate parsing, provide a title for each column in the CSV file.
- Actual file size may vary slightly due to calculations.

📘
PDF File Data Source Reminder
- The maximum file size is 20 MB.
- One language per data source is recommended.
- Images or multiple text columns won't be scraped.
- Encryption or password-protected files are not supported.
- Actual file size may vary slightly due to calculations.

📘
URL Data Source Reminder
- Include the protocol, domain, and path (if applicable) for the URL format. To effectively narrow down the data scope, it is recommended to specify the path in the following format: https://docs.example.com/docs.
- The data source will include only the data under this specific domain and path. The data from other domains or different paths will be excluded.
- One language per data source is recommended.
- Images or multiple text columns won't be scraped.
- Use a valid public URL. For private sites, export to PDF or MD files instead.
- A server error or anti-crawling may lead to a failed status during data processing.

📘
Markdown Data Source Reminder
- You can create a markdown file using any text editor or Notepad and save it with a .md extension.
- The maximum file size is 20 MB.
- One language per data source is recommended.
- Images or multiple text columns won't be scraped.
- Actual file size may vary slightly due to calculations.
- Encryption or password-protected files are not supported.

The table below shows the status and descriptions of the uploaded data source.
| Status | Description |
|---|---|
| Processing | CINNOX Dashboard is processing the uploaded data source. In this status, it cannot be selected as a data source for the CINNOX Q&A Bot. |
| Ready | The data source is successfully uploaded. It can be selected as a data source for the CINNOX Q&A Bot. Refer to the CINNOX Q&A Bot page for details. |
| Failed | The data source failed to upload during the uploading process. |
📘
- You can only use a data source for your chatbot with a status = "Ready".
- When the data source is successfully processed or failed to process, you will receive a notification message from the CINNOX Bot.
- If the data source status is "Failed", agents should review the reason for the failure and edit the data source file accordingly.





📘
Reminder
Only PDF, CSV, and Markdown file types of data sources are downloadable in any status.


📘
You cannot delete a data source if your chatbot is using it. Remove it from the chatbot - data source first, then try again.
📘
Please refer to CINNOX Q&A Bot Configuration for the Detailed Guide of Q&A Bot creation.
The Smart Reply feature is designed to streamline customer support by automatically generating suggested responses to chat enquiries. Powered by AI and your uploaded chatbot data sources, it ensures that replies are contextually relevant, accurate, and aligned with your brand's tone and knowledge base. This integration ensures that replies are not only fast and relevant but also grounded in your organisation’s trusted content.

