We are thrilled to introduce you to the latest enhancement in our line-up, an upgrade that poses a significant leap for our
Conversation AI Bot training feature. Are manual inputs when tailoring your bot's responses proving too time-consuming? We have the solution! You can now promptly provide the bot with either a single or multiple URL(s), be it for a website or blog, eliminating any extra manual work. As a result, whenever a question arises, your bot will seamlessly extract the right answers from the pool of information you've supplied.
In order to locate this feature, you just need to look in the "Bot Training" tab within the "Conversation AI" settings page, which is found in the sub-account settings sidebar.
There are three different modes available when it comes to training your bot.
By choosing the "Exact URL" option, you are opting for recommended precision training. The bot will crawl the exact URL that you've provided, using this to train itself based on what information it finds there. It's simple to use; navigate to the "Bot Training" in Conversation AI Bot's settings, choose the option "Exact URL," then enter the URL you want to crawl and hit "Get Data."
The "All URLs in this Domain" option enables a wider range of training information to be obtained, all from a specific domain. The bot will crawl through all pages and links on your specified domain, giving you the choice to select which pages the bot learns from.
To use this function, go back to the "Bot Training" in Conversation AI Bot's settings, choose the option "All URLs in this domain," enter the URL, and hit "Get Data." Be patient and wait for the pages to load, and then you'll be presented with a list of available URLs. From this list, choose the pages that will be beneficial for training the bot and hit "Train Bot."
When it comes to choosing which pages to use for training, you'll be presented with two lists: "New Pages," which are URLs not currently part of the bot's existing training data, and "Existing Pages," which are URLs already part of the bot's training dataset, visible in the "Uploaded Links" table.
By choosing to utilize the "All URLs with this Path" function, your bot will crawl all pages on the provided URL enabling you to select training pages, based on their specified path within the page URL. To use this function, you follow a similar process to before: provide the URL, hit "Get Data," wait for the pages to load, and then choose the pages for your bot to learn from before hitting "Train Bot."
With these radical improvements to our Conversation AI bot's learning capabilities, bear in mind that the URLs are not instantaneously added to the "Uploaded Links" table as bot training is a time-consuming process. Please wait for all URLs to appear in the table before testing the Conversation AI bot.
Once your bot has been trained, links/URLs can be viewed in the "Uploaded Link" table. Each URL will fall into one of two statuses: "Trained," showing the bot learned successfully from this URL, or "Failed," which reveals that the bot was unable to learn from this URL.
When refreshing a URL, remember to wait for the last data refresh timestamp to update before testing the bot.
While there's no limit to the number of URLs your bot can be trained on, precise and relevant data boosts your bot's performance and responses. Regularly review and delete irrelevant URLs from the "Uploaded Link" table.
Before you bring your training to an end, ensure you save your Conversation AI bot settings by selecting a calendar before testing the bot.
That's it! Upgrade your bot training experience now, for smarter and more efficient Conversational AI responses.