Knowledge base chatbot that answers your customers from your own material

Give HiQBot your policies, manuals and price lists. The knowledge base chatbot answers on website chat, WhatsApp, Instagram, Messenger, Telegram and LINE, and passes the chat to your team when the answer is not there.

7-day free trial. Upload a file and ask it a question the same day.

NNorthfoldBusiness account
Hi, my Duo espresso machine stopped heating. I bought it 14 months ago. Is it still under warranty?9:12 PM
Yes. Duo machines carry a two-year warranty from the delivery date, so you are covered. Our warranty policy asks for the order number and a photo of the serial label on the back.9:12 PM
Order NF-48213. Can you send a courier to collect it tomorrow?9:13 PM
Collection dates are not in the information I have, so I won’t guess. I’ve passed this to our team with your order number, and Chris will reply here in the morning.9:13 PM
WhatsApp.Answered from the warranty policy, handed over where the policy stops.

What is a knowledge base chatbot?

A knowledge base chatbot is an AI chatbot that answers customers from a business’s own material, the policies, manuals, price lists and pages its team already uses, and says so when the answer is not there.

Ask a scripted chatbot something nobody wrote a reply for and it shows you a menu. A knowledge base chatbot reads the question, looks it up in what you gave it and answers in a sentence, the way a well-briefed member of staff would. Your knowledge base here is simply that material. If you want the background on how HiQBot stores and uses it, see the AI knowledge base in our FAQ. This page is about what customers get: answers at any hour on the channels they already use, and a person when the material runs out.

HiQBot is a conversational platform for customer service, with AI agents and live chat on website chat, WhatsApp, Instagram, Messenger, Telegram and LINE.

Sources

A chatbot trained on your knowledge base, in the formats you already have

Most businesses already have the answers written down somewhere: a returns policy, a manual, a spreadsheet of prices, the FAQ page nobody reads. Building a chatbot from documents like these takes an afternoon, because there is nothing to rewrite.

  • Files. PDF, Word, PowerPoint, Excel, HTML, EPUB and plain text. Upload several at once and watch each one process.
  • Pasted text. The answer your team gives every week, typed or pasted straight in, without making a document first.
  • Product lists and Google Sheets. A price list, menu, room list or catalogue as a spreadsheet or a linked Google Sheet. Rows with a picture show as cards on website chat, Telegram and LINE.
  • Website pages. One page or a whole site of up to 50 pages, so your shipping, returns and FAQ pages become answers too.

One knowledge base serves every channel, so WhatsApp and your website never give different answers.

How answers stay inside your material

The chatbot answers from two things: your knowledge base and the conversation in front of it. When Jenna asks whether her coffee machine is still under warranty, it finds the warranty terms in your policy and answers with them. When she asks for a courier date that is in none of your documents, it tells her it does not have that information and brings in your team. It does not read the customer’s other chats or guess from general knowledge about coffee machines.

Rule-based chatbotKnowledge base chatbot
How it is builtSomeone writes every question, keyword and replyYou give it the material you already have
A question nobody scriptedFalls back to a menu or "I didn’t understand"Answers if the material covers it, says so if it does not
When a policy changesEvery affected flow has to be editedReplace the document or update the sheet
What you learn from itWhich buttons people pressedThe questions customers asked that your material did not answer

Handover

When the answer isn't in your material

No knowledge base covers everything. What matters is what happens next. For each channel you choose one of two routes: the chatbot hands the conversation to your team, who see everything said so far and reply in the same chat, or it takes the customer’s details so someone follows up later. Either way it tells the customer what is happening, and it can ask for a name, email or phone first.

Every question it could not answer lands in your knowledge gaps list, with how often it was asked. Answer it once from the dashboard and it goes into the knowledge base, so the next customer gets the answer. Where a problem needs fixing rather than answering, support tickets can pick it up.

Knowledge base chatbot use cases

A knowledge base AI chatbot for customer service earns its keep wherever the same questions come in every day and the answers already exist in a document. HiQBot’s AI agents come tuned for each trade, so the answer can turn into a booking, an order or a lead.

Online stores

Returns windows, warranty terms, sizing and delivery times from your policies, with product answers from your catalogue.

Hotels and guesthouses

Check-in times, parking, pets and what is in each room, answered at midnight from your house rules.

Software companies

Setup and how-to questions from your help docs and release notes, with a ticket when something is broken.

Tour operators and travel agencies

What a tour includes, visa and entry notes, and cancellation terms, only from the documents you upload.

Real estate

Listing details, fees and area notes from a listings sheet, with a viewing booked when the buyer is ready.

Restaurants and cafés

Menu, allergens and opening hours from your own menu file, before the table request or the order.

AI chatbot for knowledge base: features to look for

Most knowledge base chatbots look the same in a demo, answering a question the vendor chose. These are the things that decide how one behaves with your customers on a busy Monday.

  • Answers only from the material you give it, and says so when the answer is missing
  • Takes files, pasted text, Google Sheets and website pages, not one format only
  • Hands the chat to a person who sees the whole thread, or takes details for a follow-up
  • Shows you the questions it could not answer, ranked by how often they came up
  • Works on WhatsApp, Instagram and the rest from the same knowledge base
  • Replies in the language the customer writes in
  • Can do something with the answer, such as book an appointment, take a lead or open a ticket

One knowledge base, six channels

Customers ask the same question in different places. The chatbot answers on all six from the same material, in the language the customer writes in, and every conversation lands in one inbox. Customers can also send a photo, a product label or a damaged part, and the AI reads it.

Set it up in an afternoon, keep it current every week

  1. 1

    Gather what you already have

    Return and warranty policies, product manuals, price lists, the FAQ page on your site. The documents your team already looks things up in are the right place to start.

  2. 2

    Add it to your knowledge base

    Upload the files, paste the odd answer, link the Google Sheet and add your website pages. One knowledge base serves every channel.

  3. 3

    Choose what happens when it does not know

    Hand the chat to your team, or take the customer’s details so someone follows up. Pick which details it collects first.

  4. 4

    Switch on your channels

    Website chat, WhatsApp, Instagram, Messenger, Telegram and LINE, or only the ones your customers use.

  5. 5

    Fill the gaps every week

    Open the knowledge gaps list, answer the questions it could not, and the next customer who asks gets the answer.

What it will not do

    Does not answer from Notion, Google Drive or Confluence
    Does not publish a help center for customers to browse
    Does not invent an answer your material does not support
    Does not write anything back to your Google Sheet
    Does not make or answer phone calls
    Does not let you pick the AI model or bring your own key

Questions about knowledge base chatbots

Last updated

What is a knowledge base chatbot?

A knowledge base chatbot is an AI chatbot that answers your customers from your own material: policies, manuals, price lists, product sheets and website pages. HiQBot answers on website chat, WhatsApp, Instagram, Messenger, Telegram and LINE, and hands the chat to your team when the answer is not in what you gave it.

What can I train it on?

Files such as PDF, Word, PowerPoint, Excel, HTML, EPUB and plain text, text you paste in, product lists, a linked Google Sheet, and pages from your website, one page or a whole site of up to 50 pages. Notion, Google Drive and Confluence are not connected.

Does it only answer from my documents?

It answers from your material and from the conversation in front of it. When the answer is not in your material, it says so instead of guessing, and the question is added to your knowledge gaps list so you can fill it in.

What happens when it does not know the answer?

You choose per channel. It can hand the chat to your team, who see everything said so far, or take the customer’s details so someone follows up later, and it tells the customer which one is happening. It can ask for a name, email or phone first.

Do I need to write code or build a template?

No. Upload files, paste text, link a Google Sheet or add your website pages in the dashboard, and the chatbot starts answering from them. You can upload several files at once and watch each one process.

How do I keep it up to date?

Replace a file, edit the text or update the sheet. A linked Google Sheet refreshes on a schedule and when you press sync. The knowledge gaps list shows the questions it could not answer and how often each came up, and you answer them from the dashboard.

Is it the same as help center software?

No. HiQBot does not publish a help center or article pages for customers to browse. It reads your material and answers customers in chat. If you already have a help center, add its pages as a source.

Which channels and languages does it work in?

Website chat, WhatsApp, Instagram, Messenger, Telegram and LINE, from one knowledge base and one inbox. It replies in the language the customer writes in, with nothing to set up.

Is my data used to train AI models?

No. Your customer data is never used to train models, each organization’s data is kept separate, and data is encrypted in transit and at rest. Email addresses and phone numbers are masked in the text the AI reads.

Can customers send a photo instead of describing the problem?

Yes. On website chat, WhatsApp, Instagram, Messenger, Telegram and LINE the AI reads photos a customer sends, such as a product label or a damaged part, and answers from your material.

Upload one policy. Ask it the question your team answers most.

Start with the document your customers ask about most and see the answer arrive on WhatsApp before you pay for anything.