What an AI knowledge base is for customer chats
An AI knowledge base is a collection of the business’s own materials, prices, services, delivery terms, policies and frequent answers, that AI reads before replying to a customer. It is different from a general AI chatbot that answers from anything it was trained on: here the source is limited to what the business itself put in, so the answer matches reality.
Deciding where AI should answer and where a person should
Not every channel needs the same level of automation. A business can let AI send replies on its own in the website chat, where questions are usually simple, and keep WhatsApp as draft-only, where a person reviews before sending. This choice is set per channel in the inbox, not all or nothing.
Keeping prices and policies in one place instead of five
The most common cause of a wrong answer is not AI, it is an outdated price sitting in someone’s notes while the real one changed weeks ago. A single knowledge base removes that gap: update the price once, and every answer, in the inbox and inside chat flows, follows the same update.
What happens when a question is outside the knowledge base
When a question needs a judgment call, a refund exception, a custom order, a complaint, the conversation moves to a team member instead of AI guessing. Nothing about the exchange is lost: it stays in the contact’s history together with everything answered automatically before it.