First, head over to BotSailor's Dashboard, Go to Integrations, & click, 'AI Training Campaign'. From here, then click, 'Create', to train a new AI campaign. You can also edit & update older trained campaigns from the Action menu.
Let’s Set Up the Primary Prompt for the AI Chatbot: After clicking new, this page will open where you can add the 'Campaign Name' & 'Prompt Message' .
There is already a default Prompt message for the bot. It's an initial instruction to the bot to help it understand its role. You can tone it according to your needs & the AI chatbot online will know its functions. After naming the campaign, let’s update the default prompt and give more precise instructions. For this, We have added,
'Your responsibility is to support our customers as an agent or representative for our company. Always represent yourself as "we" or "our company" when discussing features, services, or policies. Your role is to assist customers with accurate and helpful information about our offerings. When addressing customer inquiries about prices, provide clear and concise information, including any available discounts, packages, or subscription options. Keep your responses professional, friendly, and informative & ensure that the customer understands how our products and services can best serve them.
Training the AI Chatbot With Manual FAQ:
After the Campaign is created, let's modify it by clicking the Plus button from the 'Actions' menu. You can train the AI either with a summary or an FAQ. While a summary is more intelligent as context is available on your business as a whole, richer responses will be issued. In line with this, a summary consumes more tokens than an FAQ which is a more cost efficient format that uses specific questions and answers. Simply upload your content in the required format, save the settings, and your AI is good to go. For richer, more complex responses use a summary or if promptness is what you seek, you can rely on FAQs.
Then we have added our business summary here for better response. You can add as many FAQ’s as you like or add a summary.
That's all. Now, the chatbot will do its wonders & reply from the given prompt as a summary.
Training the AI Chatbot With URL:
Now, insert an URL into this open AI chatbot setting to fetch data! First, switch to the URL from the right corner and click 'New.'
Following that, this page will open, where you have to add the Campaign URL, of the page from where the bot will fetch data, in the 'Fetch Content Configuration' section, you must select a 'Selector,' which can be either 'ID' or 'Class.' Well, this depends on the website's design. Some elements on the page are defined by an ID (unique to that element), while others are defined by a Class (which can apply to multiple elements). The choice of Selector will depend on the website's structure and which one best targets the specific content you wish to fetch.
You can also Remove Content Configuration by ID or class to specify areas to exclude if the webpage contains unnecessary details, such as ads, navigation bars or footers, irrelevant to the content you want to fetch. Removing these sections ensures that only the necessary content is included in the fetch process.
Now after adding the URL let’s complete the process. Here, we will add BotSailor's landing page link. We specifically selected the part where BotSailor's Integrations are mentioned, and after inspecting it, I can see it's under the 'Class' section.
From here, We copied the selector name, 'Container.'
Now, we will apply the input in the Selector and Selector Name section. The Selector is 'Class,' and the selector name is 'Container.'
Since we only selected one part of the page to fetch data, we do not need to remove any content configuration. Finally, once done adding the Class & Selector Name, click 'Generate FAQ' or 'Generate Raw Response' & save. That's all you will be good to go. That’s all about the URL input, you can either give specific ID or Class from a selected section or the entire page. In that case you should remove unnecessary content by using the removing content configuration option, & remove headers or footers.
Training the AI chatbot with File:
You can upload structured documents like training manuals, FAQs or product guides into your chatbot from the File Settings section. Click 'New' to upload your PDF, Word (,doc TXT or) file.
Once you upload your file, you have 2 processing choices :
Generate Raw Response: This option returns the whole document as an individual, comprehensive response. By giving the whole raw response in one FAQ, it provides a far far more in depth answer and also allows the chatbot to react in a more context. However, this particular method requires more tokens and can use up far more credits because the responses are detailed.
Generate FAQ: This option divides the content into smaller, individual FAQ entries. This approach utilizes fewer tokens and it is less costly, however the chatbot might not offer as context-aware or detailed responses as it would with a raw response. However it makes for quicker, much more succinct answers.
When finished, select your choice and click 'Save'. The chatbot will then make use of the uploaded file to offer responses - in-depth or concise - per method chosen.
Setting Up the No Match Reply & Configuration
Since we have completed training the free AI chatbot online with necessary URLs, files, & Manual FAQs as needed. Let's configure the No Match Reply, where the bot will reply to any customer query according to the training data upon no match!
It will redirect you to the visual drag-and-drop flow builder. Double click ‘AI Reply’. In the 'Open AI training campaign' section, select and save the trained bot.
In the next step, scroll down in the Bot Manager & click 'Configuration.'
Enable the 'No Match Reply' & save settings. No match reply setting with AI Campaign is the best option to make your open AI chatbot active all time.
Configuring the Central Control System: "Enable AI Assistant"
Simply Navigate to the Bot Manager. Locate the "Enable AI Assistant" & switch it on, besides this the chatbot menu will appear that are already trained, you can select your preferred campaign. From here you can control your chatbots behavior, to learn the controlling, prefer reading this blog- AI Assistant Chatbots: Revolutionize Your Customer Support .
The intent detection in the AI Assistant system is a whole new part, to learn how you can set the intent detection, go here & read- AI-Powered Intent Detection to Enhance Chatbot Efficiency .
Now set the chatbot to either ‘AI Assistant' for All Queries’ or ‘AI as a Fallback Only’ according to your need.
Let's test the bot now from the user end!
Cleary, the open AI chatbot is replying perfectly. Your free AI chatbot is now fully trained and configured. Now this bot can respond to your customer inquiries using the trained data from manual FAQs, URLs and uploaded files. When "No Match Reply" or AI Assistant for ‘All Queries is enabled’, the bot can provide helpful answers, which will ensure smooth customer support.
Related Blogs:
AI Assistant Chatbots: Revolutionize Your Customer Support
Train AI Assistant for Chatbot With FAQ, URL & File
AI-Powered Intent Detection to Enhance Chatbot Efficiency
Video Tutorials:
How to Create a Customer Support AI Chatbot
How to Train Your AI Assistant Using FAQ, Files, and URLs
How to Create an AI-Integrated No-Match Reply for WhatsApp on BotSailor
1. What is the difference between FAQ and Raw Response training methods?
FAQs are token-efficient and suitable for quick replies. Raw Response enables deep, contextual answers but uses more tokens and credits.
2. Can I combine all three methods: FAQ, URL, and File?
Yes! BotSailor allows you to combine FAQs, URLs, and uploaded files in a single training campaign to enrich your chatbot's knowledge base.
3. What types of files can I upload to train the chatbot?
You can upload .pdf, .doc, and .txt files, such as product guides, help manuals, or onboarding documents.
4. How do I fetch content from a webpage using URL training?
Paste the page URL and select a content area using CSS Selectors (ID or Class). You can also remove unwanted elements like headers and footers.
5. Is AI training reusable across multiple bots in BotSailor?
No, training campaigns are bot-specific. You'll need to create and configure separate campaigns for each bot.
6. What happens if the chatbot doesn’t understand a query?
You can configure a “No Match Reply” using AI training. The bot will then fall back to your trained data to respond meaningfully.
7. How do I update an existing AI campaign?
Go to Integrations > AI Training Campaign, find your campaign in the list, and use the Action menu to edit or update it.
8. How can I test if my AI chatbot is working correctly?
Once your training is saved and activated, test it by sending real customer-like queries through your bot interface and observe the responses.
9. Can I control when the AI should respond vs. fallback to default flows?
Yes. Use the setting in Bot Manager > Enable AI Assistant to toggle between “AI for All Queries” or “AI as Fallback Only.”
10. Where can I manage Intent Detection settings?
Intent Detection is configured separately within the AI Assistant panel. Visit the “AI Intent Detection” documentation for full setup steps.