Prerequisites Make sure your account is on the Pro AI plan and you have enough AI chatbot credits available in your onWebChat account.
New to onWebChat? Create your account to get started. Get the Pro AI plan free for your first month, plus 50 AI chatbot credits (each credit covers one full chat dialog with a visitor).
Step 1 Open AI Chatbot Settings
After you log in to your onWebChat dashboard, go to the AI Chatbot section and click Website Settings. This page controls the core chatbot setup, including your website details, response behavior, and global instructions that define how your bot communicates with visitors.
Step 2 Add Website Details and Bot Instructions
Enter the following information:
Website URL: Enter the URL of your website.
Website Description: Add a short, clear summary of your business (for example, "Online clothing store").
Off-Topic Reply: This is the fallback response when the bot does not understand a question or when the request is outside your business scope.
Missing Context Reply: This is the fallback response when the visitor asks about your website/services, but the answer is not found in your training context.
Bot general guidance: Define how your AI bot should behave and communicate. These instructions apply to every response. For example: "Be friendly and professional", "Keep answers concise", or "Share relevant links when helpful". Strong guidance keeps your brand voice consistent across all conversations.
Step 3 Train Your AI Chatbot
Training is the most critical step in building a high-performing AI chatbot. The quality of your training data directly impacts how well your bot understands your business and answers customer questions. Open the AI Bot Training Data tab and add the content your chatbot should use in real conversations.
Best Practice: Add content for as many likely customer questions as possible. Use plain text or structured Q&A pairs. Include product details, service information, policy answers, FAQs, and any business-specific information your customers regularly request.
You have multiple options for training your AI chatbot:
- Manual text entry: Add training data directly by typing or pasting text into the interface (recommended method for best results)
- File upload: Upload documents, PDFs, or text files containing your business information
- Website crawling: Automatically crawl your entire website or specific pages to extract content for training
For detailed information about each training method and best practices, read our guide to chatbot training.
Step 4 Test Before You Go Live
Before publishing your AI chatbot on your live site, run comprehensive tests to verify response quality and reliability. Testing helps you spot gaps in training data, confirm your global guidance is applied correctly, and validate that the bot handles various question types effectively.
Start several test conversations covering different customer intents and edge cases. Review each response for accuracy, tone, and relevance. Test out-of-scope questions to ensure your fallback message works as expected. Then, refine your training data and bot instructions based on your findings.
Step 5 Monitor and Make Continuous Improvements
Once your AI chatbot is live, continuous optimization is key to long-term success. Review conversation logs in your onWebChat dashboard to measure quality, identify common customer questions, and uncover opportunities to improve automation.
If your chatbot misses context or provides weak answers, take these steps:
- Add new training data for repeated customer questions
- Refine your global guidance to better match your brand voice
- Update your fallback message for out-of-scope requests
Continuous improvements ensure your chatbot remains accurate, helpful, and aligned with your growing business.