---
title: 'How to Build an AI ChatBot with Your Own Knowledge Base in 10 Minutes (Coze Tutorial)'
source: 'https://youtube.com/watch?v=_EbPN38gE_w'
video_id: '_EbPN38gE_w'
date: 2026-06-17
duration_sec: 0
---

# How to Build an AI ChatBot with Your Own Knowledge Base in 10 Minutes (Coze Tutorial)

> Source: [How to Build an AI ChatBot with Your Own Knowledge Base in 10 Minutes (Coze Tutorial)](https://youtube.com/watch?v=_EbPN38gE_w)

## Summary

This video demonstrates how to create a custom knowledge AI chatbot in 5-10 minutes using Coze, a free platform built on GPT-4. By uploading personal data such as YouTube scripts and ebooks, users can train the chatbot to provide personalized advice without any coding experience.

### Key Points

- **Introduction to Coze** [00:00] — Coze is a free platform that runs on GPT-4, offering a cost-free alternative to OpenAI's $20/month subscription.
- **No coding required** [00:47] — Building AI chatbots on Coze requires no programming skills, making it accessible to beginners.
- **Creating a bot** [01:30] — Go to coze.com, create an account, and click the 'Create Bot' button to start.
- **Designing the persona** [02:24] — Define the bot's behavior through prompt engineering, then use Coze's 'optimize' feature to refine it.
- **Uploading knowledge** [04:21] — Add data sources (PDF, txt, Word, Google Drive, Notion, web pages) by creating a knowledge source and uploading documents.
- **Testing the bot** [07:56] — Before uploading knowledge, the bot gave generic GPT-4 responses; after upload, it provided answers based on the creator's own content.
- **Adding plugins** [09:38] — Plugins like YouTube search extend the bot's capabilities beyond native GPT-4 functions.
- **Publishing the bot** [10:38] — The bot can be published to Coze's bot store, Slack, or Discord with one click.

### Conclusion

The video shows a straightforward, no-code method to build a personalized chatbot using your own data on Coze, with added plugins and simple deployment across platforms.

## Transcript

in this video I'm going to take you
through creating your own custom
knowledge AI chatbot in 5 to 10 minutes
that's trained on your own data using a
platform known as Co now complete
transparency this video is sponsored by
Co however it is a completely free
platform that runs on top of GPT 4 now
if you wanted to use GPT 4 through open
ai's website you'd have to pay like 20
bucks a month but here on Co you get to
use it for free on top of all the
additional chatbot Integrations that
this platform provides now I've been
using this platform for a couple weeks
at this point and I think it is super
cool and I know a lot of you guys are
trying to learn Ai and one of the best
ways to learn AI is to sort of get your
hands dirty and start building stuff now
with Coes it's super easy to start
building with AI even if you have no
coding experience so I've had this idea
for a while now about making a chatbot
that uses my own content you know my
YouTube video scripts my ebook my
courses so that users can interact with
this chatbot to get advice based around
the content that I've made and the
things that I've said and aside from my
own knowledge chatbot I know a lot of
people or a lot of companies even could
benefit by creating their own chatbot
from their own data sources so with that
being said let's go ahead and dive into
creating this chatbot so head on over to
co.com and once you create an account
click the create bot button here in the
top left of the screen select your
personal workspace give the bot a name
so I'll name this uh coding with Kenny
we'll give the bot a brief description
so I'll just say a knowledge bot trained
on Kenny Gunman's
content then we can upload a profile
picture or we can use this handy
generate AI button right here to
generate an icon for our bot so this
will take a second to
load once that's done we'll click
confirm and then from here we'll be
taken to our bot dashboard where we can
now create our bot so the very first
thing that we should do is design the
Persona or I guess prompt engineer the
bot so it behaves the way that we want
it to for example I want this bot to
give the user coding advice from the
knowledge that I upload which we'll do
here in a second and recommended content
from my YouTube channel so I could write
something in the Persona here like
you're a coding bot designed to
recommend coding resources and YouTube
videos from Kenny
G's YouTube
channel you're able to help Rec
recommend resources based on an
individual's skill level so I would say
this is a good start for our Bots
persona but then I could go ahead and
use CO's optimize feature right here
click this
button this will take a minute or so to
run and as you see Co went ahead and
took our initial prompt and made it into
something that is much more optimized it
created a character for our bot which is
denoted by this character hashtag here
it equipped it with all of these
different skill sets such as diagnosing
the user's coding skill level
recommending coding resources and then
recommending YouTube videos from my
YouTu YouTube channel right here so I'll
hit use to go ahead and use this
generated data now I already created a
Persona and prompt section for this bot
that works very well for my use case
right here so I'm going to go ahead and
copy and paste it in like this now the
next thing that we need to do is upload
our knowledge so right now this bot
isn't trained on anything and it's
strictly the gp4 model so if I ask it
some sort of question it will give me a
generic AI response for example if I
asked it something like what are some
tips for
selftaught programmers it'll take a
second and then it's going to go ahead
and give me this pretty generic response
of mastering the basics practicing
regularly work on real world projects so
let's go ahead and upload my own data
just to see how much this response
changes so I have over a 100 pages of
PDF data from all the content that I've
created over the years for example
inside of learning material here I have
this 19-page ebook that I created titled
02 full-time programmer in five steps
inside of YouTube video scripts here I
have 10 YouTube video scripts that I've
downloaded and I've turned into PDF
documents so I'm going to go ahead and
upload all of these documents directly
into Coes and integrate them into my bot
so I can do that by going back to the
Coes dashboard clicking this plus button
under the knowledge section now I've
already uploaded these documents as you
see right here while I was testing this
bot but let me go ahead and show you how
you can add your own documents so what
you want to do is click this create
knowledge button right here create a
name for your knowledge source so I'll
name this one YouTube vid Scripts add a
description if you want click the
confirm button now this is where you
will upload your knowledge AKA your
documents and you do that by clicking
this add unit button right here now I'll
personally be using PDF document but as
you can see you can use virtually any
type of document you want you can use
text documents you can use Microsoft
Word documents you can even connect
right to your Google doc drive you can
connect to notion or you could even
scrape data from web pages but because
I'm using local PDF documents on my
machine I'll go ahead and click local
documents click next and then alls I
need to do is drag and drop the
documents that I have right here so I go
into YouTube video scripts I'll select
all of these and I will drag them
straight into codes like this they
should upload pretty quickly from here
I'll click the next button now this is
where you can let codes automatically
parse through your document or you can
set up custom rules for parsing yourself
I think Co does a great job doing this
automatically so I'll go ahead and leave
automatic selected I'll click next and
then we simply just wait it should only
take a minute or so and once that is
done we'll click the confirm button and
after that simply go back to knowledge
and upload all of the other data sources
that you have from here we will simply
go back to our bot dashboard so this
coding with kennybot right here click
the plus button under the knowledge
section We'll add our YouTube vid script
knowledge data source that we just
uploaded we'll also add newsletters and
then learning material here we'll close
out of that and just like that we've
created a knowledge bot that is ran on
top of gp4 with our own custom data and
just to show you that this is working
let's go ahead and ask the coot the same
question of
what are some tips for self toop
programmers so I'm going to go ahead and
copy this with contrl C I'll paste it in
right here I'll ask the bot again what
are some tips for self-taught
programmers and as you see it is
searching through our
knowledge and providing us with a
completely different answer than the
first time so let's look through this
response here now these things are
things that I've said many times before
in YouTube videos which makes sense
because I uploaded my YouTube video
scripts directly to the knowledge
section of this bot and just to prove
this to you let me see if I can find the
video where I said self-learning isn't
the easier route all right so I said it
right here in this video self toop
programmers listen up and as you see in
this chapter I say selftaught isn't the
easier option we'll go ahead and play a
few seconds of this clip you need to
understand that being selftaught isn't
the easier route by any means I think a
lot of people there you go directly from
one of my YouTube video scripts so that
my friend is how you can very easily
create a knowledge bot that is trained
on your own data now with Co it doesn't
stop there as I mentioned I want to
create a knowledge bot that not only
gives advice based on the data that I've
uploaded to the bot but also recommends
YouTube videos from my own channel now
out of the box gp4 can't recommend
YouTube videos however this is where the
plethora of other helpful features of Co
comes into play now Co has a huge amount
of plugins that I can add directly into
my bot to get it to do additional tasks
such as searching for YouTube videos so
I'll head over to this plugins section
in my bot dashboard here I'll click the
plus button to add a plugin so I'll go
to the search bar here and I'll search
for the YouTube plugin I'll select it
right here and then I'll add the search
video plugin directly from the YouTube
plugin so now that this has been added
we can close out of this here and now my
bot can search for YouTube videos it's
as easy as that so if I type in
something to my bot like can you provide
me with a video on good coding habits
that I can adopt as a programmer I hit
enter it will use the YouTube plugin and
then it will recommend a video directly
from my YouTube channel so now that we
have our knowledge-based chatbot created
we are ready to deploy that and again
that is something that is super easy
with Co we simply click the publish
button in the top right of the screen
here Co will ask for a few more pieces
of data such as a uh bot opener and
opening questions that a user ask you
can fill this out if you want or you can
skip it I'll just hit confirm to go with
the default it gives
me and then from here we can deploy our
bot to various different locations so
the first one and the default one is
directly to the co bot store So Co has a
bot store where you can deploy your Bot
to and you can use Bots that were
deployed by other people but you can
also integrate with platforms such as
slack and Discord so you could install
this directly into your Discord server
and the users of your server could ask
the bot questions and it would answer it
so I think that is a really cool feature
however I'm just going to publish this
directly to the co bot store for now
just because that is the easier route
and I do that by clicking this publish
button right here and there we have it
our bot is now deployed so I'll click
this button right here to open up the
bot in the bot store and that my friends
is how you create a knowledge chatbot
using Coes I hope this video was helpful
to you and make sure you head on over to
co to start building your own chat Bots
today thank you for watching the video
and I will see you all in the next one
peace