Test Your AI Agent
Before publishing your AI agent, it's important to test how it responds in real conversations. The AI Testing page provides a live preview where you can chat with your assistant, verify its responses, review the information it uses, and ensure everything is working as expected before your customers interact with it.
Introduction
After configuring your AI agent and building its Knowledge Base, the final step is testing. This allows you to experience your AI exactly as your visitors will, helping you identify anything that needs to be adjusted before deployment.
The AI Testing page provides a complete preview of your assistant in a safe testing environment. You can ask questions, try different scenarios, and confirm that your AI responds accurately using the information you've provided.
Chat with Your AI
The main area of the page displays a live chat interface where you can interact directly with your AI assistant. Ask the same questions your customers are likely to ask and verify that the responses are accurate, natural, and aligned with your business.
Testing different conversation scenarios is one of the best ways to evaluate your AI. Try asking about your services, pricing, business hours, policies, or frequently asked questions to confirm that the assistant understands your content and responds appropriately.
Review Your AI Configuration
Alongside the chat preview, you'll find a summary of your AI agent's configuration. This includes the selected AI model, response style, language, temperature, response length, and other important settings. Reviewing this information helps you quickly confirm that you're testing the correct configuration.
You'll also see an overview of your AI's available knowledge, including indexed knowledge sources, FAQs, uploaded files, website pages, services, business availability, and any enabled features. This provides a quick snapshot of the information your AI has access to during conversations.
AI Context Log
One of the most useful tools on the AI Testing page is the AI Context Log. This allows you to inspect the context that was sent to the AI for each response, making it much easier to understand how the assistant reached its answer.
The context log is especially helpful when troubleshooting unexpected responses. If the AI misses information or provides an incorrect answer, you can review the context being supplied and identify whether the issue is related to your Knowledge Base, AI instructions, or configuration settings.
Refine and Test Again
Testing is an iterative process. If you notice that your AI doesn't respond the way you expect, simply update your AI instructions, improve your Knowledge Base, adjust your settings, or add additional information. You can immediately return to the testing page and continue chatting with your AI until you're satisfied with the results.
Spending a few extra minutes testing your assistant before publishing it can significantly improve the experience your customers receive.