When an AI agent needs to answer a question, it searches the knowledge base for relevant content and uses what it finds to form a response. Without a knowledge base, the AI falls back on its training data, which may be outdated or wrong for your specific business. With a knowledge base, the AI answers from your actual documents.
You can populate a knowledge base with product documentation, pricing FAQs, service descriptions, onboarding guides, or anything else clients commonly ask about. GoHighLevel indexes the content so the AI can search it quickly during a live conversation.
Knowledge bases reduce hallucinations. An AI that says "I don't know" or gives a wrong price because it was not trained on your rates is a problem. Grounding the AI in your own documents keeps answers accurate and on-brand.
How Knowledge Base works
You create a knowledge base in the sub-account AI settings. You upload documents (PDF, text, web pages) or type content directly. GoHighLevel chunks and indexes the content. When you attach this knowledge base to a Conversation AI agent or an Agent Studio agent, the agent searches it every time it needs to answer a factual question.
The search is semantic, meaning the agent finds the right document even if the contact asks the question in a different way than you phrased the answer. The agent then synthesizes a reply from the retrieved content and the conversation context.