- Navigate to the 'Sources' page, and select 'Scrape Zendesk'.
- On the 'Zendesk' page, click on 'Scrape new zendesk'.
- A dialog box will appear enter a zendesk URL directly and click submit. The platform will start scraping all articles from your Zendesk knowledge base.
- After the scraping process is complete, the platform will display all scraped articles in a table. This table will show details such as the article URL, the status, and the last trained date.
<aside>
💡 Note: The scraping process might take some time depending on the number of articles present in your Zendesk knowledge base. The more extensive the knowledge base, the longer it will take to scrape.
</aside>
Once scraping done, the platform will list down the links in a table. This table includes columns for the Article, character count, the last trained on, and status.
- Article: The name of the article
- Character count: The total number of characters scraped from the article
- Last trained on: This will be 'Empty' for a new article, but will show the date when the bot was last trained using this article for existing articles.
- Status: Indicates the current state of each individual source and its involvement in the bot training process.
- Actions: These actions are used to manage the training status of the article
- Enable/Disable:
- 'Enable' will appear if the article status is 'Skipped', clicking this will change the status to 'Draft'. '
- Disable' will be shown if the article is in 'Draft' mode or 'Trained' mode. By clicking this, you can remove the article from the bot's training process.
- Refresh: This action will appear if the bot status is 'Trained'. By clicking 'Refresh', the platform will rescrape the specific article and retrain the bot with any new content from that article. This allows your bot to stay updated with the latest information from the source.
<aside>
💡 Remember, you can add as many zendesk urls as you like until you reach the character limit set by your chosen plan. With each new article, your bot gains a wider knowledge base to pull from, enhancing its ability to assist your users
</aside>
How status works?
Each source used for training your bot displays a status indicating its current state in the training process. This guide will help you understand what each status means, enabling you to manage and monitor your sources effectively.
Here's a breakdown of what each status signifies:
- Draft: If a source's status is marked as 'Draft', it signifies that no action has been taken on this source yet. It remains unprocessed and has not been utilized in the training of your bot.
- Skipped: When a source's status is 'Skipped', it indicates that the source has been deliberately overlooked and will not be used for bot training. You might choose to skip certain sources that are irrelevant to the bot's knowledge base or that contain unnecessary information.
- Trained: If a source's status is 'Trained', it means the source has been used to train your bot. The data or information contained in these sources has been integrated into your bot's knowledge and can be used to assist your users.
- Training/Retraining: A source status of 'Training' or 'Retraining' implies that the source is currently being used for the bot's training process. The bot is actively learning from the information provided in these sources.
<aside>
💡 Note The 'Skipped' status is only applicable at the domain page level and not for each individual source.
</aside>
Knowing what each status means is crucial for understanding your bot's training process. This knowledge provides insights into what your bot has learned, what it is currently learning, and what it will potentially learn in the future.