Skip to main content
Knowledge base

Whatever the bot knows, it learned from you.

You hand over your website as a link and your catalogue as a file. From then on every question about prices, terms, sizes and rules is answered from that material — in a chat or on a phone call.

  1. Material goes in
  2. The base is built
  3. Answers come from it
  4. Changes apply at once
The problem

The bot either doesn't know, or knows the old answer.

Most chatbot projects stall here — not on the technology, on the material.

  1. 01

    You're asked to write it from scratch

    “Write 200 questions and answers to train the bot” is usually where the project stops. Those answers already exist — on your site and in your message history.

  2. 02

    The price changes; the bot doesn't

    A campaign starts and the bot is still quoting last month. Nobody notices until somebody complains.

  3. 03

    The same material, kept twice

    One base for the chatbot, another for the call agent. The day they drift apart, a customer gets two different answers.

The answer to all three is the same principle: the base is built from material you already have, it is kept in one place, and a change takes effect across every channel at once.

Sources

Five ways in. None of them starts with a blank page.

The more material you already have, the less work is left.

  1. You give us your website

    Pages are crawled automatically, the text is cleaned and turned into material — up to 4,500 pages. Your service descriptions, your terms and your FAQ are already written there.

  2. You upload PDFs

    A price list, a technical document, a contract template — files up to 25 MB are turned into text and added to the base.

  3. You upload your catalogue

    Name, code, price, stock, categories and variants — each size and colour with its own price and stock. Photos go in too, which is how a product is recognised when a customer sends a picture.

  4. It learns from past conversations

    A question-and-answer set is generated from your Instagram account's old threads. The bot starts out knowing what your customers actually ask.

  5. You fill the gaps by hand

    Anything left is added from the panel — one at a time, or in bulk, up to 200 entries per request.

Knowledge base

Answer built from the source

  1. Customer

    How much is this chair, and is it available in black?

  2. Vexvon

    The office chair is 340 AZN. It comes in black and grey; black is in stock right now.

  3. Customer

    How long is the warranty?

  4. Vexvon

    It carries a 2-year warranty. The terms are in our warranty document — a copy is handed over on delivery.

The first answer came from the product catalogue, the second from a PDF. To the customer there is no difference — both are in the same conversation.

What it can do

Four jobs on the knowledge side.

Material is gathered, organised, served to two AIs at once, and kept current. Open any of them to see what is inside.

Website crawling
You give a link, the pages are walked automatically and stored as cleaned text — up to 4,500 pages.
PDF documents
Files up to 25 MB are uploaded, their text extracted and made searchable.
Product catalogue
Stored as structured products: variants with their own price and stock, and their photos.
Instagram archive
A question-and-answer set, or a general company context, is generated from past conversations.
Bulk upload
Up to 200 entries can be added in a single request.

See how an answer is built
Scenarios

The moments the knowledge base proves itself.

All four answer the question “where did the bot get that?”

01

Day one

«How soon can the bot be ready?»

A link to your site and a catalogue file are enough. You never sit down to write a question-and-answer list.

02

A campaign starts

«We've changed the sale prices»

You update the catalogue. From that minute the bot quotes the new price — on every channel.

03

A question from a document

«What are your warranty terms?»

The answer comes out of the PDF. Nobody turned that document into Q&A by hand.

04

The bot got it wrong

«Why did it say that?»

Because the source of each answer is stored, you find the material and fix it — once, for everywhere.

The Vexvon platform

Loaded once. Works on every channel.

Instagram, WhatsApp, Telegram, Facebook, TikTok, your website and your phone line all speak from the same base.

A knowledge base belongs to a company, not to a channel. Which is why changing a price in one place is enough: whether the customer asks on Instagram or picks up the phone, they get the same answer. That removes the root cause of inconsistent replies.

Demo

Give us your website link and we'll show you the rest.

Showing it on your own site is shorter than explaining how a base gets built. Tell us briefly about your company — our AI Call Center agent will call you shortly and answer your questions.

  • From your site, PDFs, catalogue and archive
  • Shared by the chatbot and the call agent
  • A change applies on every channel at once
Book a demo

Your details are used only for the demo and to get in touch.

Book a Meeting with Vexvon

Pick a time that suits you in our calendar.