[0:00] Today, I'm going to show you how to build your own 
AI powered customer support chatbot using Zapier  
[0:06] with absolutely no coding. I made this video 
in partnership with Zapier. We'll create a bot  
[0:13] for the Kevin Cookie Company that can answer 
questions, pull from your business knowledge,  
[0:17] and even log support requests automatically. 
And at the end, I’ll show you how to embed the  
[0:23] chatbot directly on your website so customers 
can start using it right away. The best part?  
[0:29] Zapier chatbots connect to all your existing 
tools. So instead of jumping between apps,  
[0:34] your bot can actually take action, like creating 
a support ticket or sending an email. This makes  
[0:39] it incredibly powerful for real businesses. I'm 
Kevin, and let's start creating our chatbot. To  
[0:46] follow along, all you need is a Zapier account, 
and it's free to get started. I've included a link  
[0:51] in the description and as a pinned comment. Once 
you're signed in, click create on the left-hand  
[0:57] side and go down to chatbots, which is where you 
can build and customize your chatbots. You don't  
[1:03] need any coding experience. Now Zapier gives me 
two ways to create a chatbot. I can have Zapier  
[1:09] build a starting point automatically based on my 
website, or I can create one from scratch. For  
[1:14] this video, I'll choose create from scratch 
so we can walk through each step and fully  
[1:19] customize the bot. Right down here, we can give 
the bot a name. I'm going to call mine Chip,  
[1:24] the cookie helper. Once you enter a name, let's 
click on create. Now that the bot's created,  
[1:29] we can set up its personality and its behavior. On 
the left-hand side, I'm in the instructions tab.  
[1:35] This is where you tell your bot how to greet 
people, and underneath that, how it should  
[1:40] respond. For the greeting, I'll keep it simple and 
just ask, how can I help you today? Below that,  
[1:45] we have the directive box. This is the main 
set of instructions for the bot. Now, overall,  
[1:51] the default looks good, but I do need to customize 
a few things. Right up here in the objective,  
[1:55] you are an exceptional customer support 
representative. Your objective is to answer  
[1:59] questions and provide resources about, 
and you notice I need to customize this,  
[2:04] so let's enter in information about the Kevin 
Cookie Company, a business that bakes and sells  
[2:08] a variety of cookies. Now, the rest of this 
looks good. Right down here, you could also  
[2:12] specify the style of the communication for 
your bot, and at the very bottom, we have  
[2:16] other rules. This one's very important. Your 
answer must come from the information returned  
[2:22] from that knowledge source. In a moment, we'll 
contribute information to the knowledge source,  
[2:26] and it'll use that to answer questions. We 
don't want our bot making up information.
[2:32] Once all of this looks good, down at the very 
bottom, let's click on save changes. Let's now  
[2:36] add the information that the chat bot will use to 
answer customer questions. Over on the left-hand  
[2:42] side, let's click on the knowledge tab. This is 
where you can connect documents, web pages, files,  
[2:47] and even Zapier tables. Let's start by adding 
our first knowledge source. Right up on top,  
[2:52] I'll click on this button. This opens up a 
dialog, and right over here, we can select the  
[2:56] first source. I'll start with the website. I'll 
select this, and right down below, I'll type in  
[3:01] the website for the Kevin Cookie Company. Down 
below, I can also crawl this link for all the  
[3:05] subpages. I'll check that box, and then right over 
here, let's click on continue. It looks like it  
[3:10] found a number of different subpages. Let's check 
the ones that we would like to include as part of  
[3:14] the knowledge source. I’ll go with recipes, 
here we have chocolate chip, and let's also  
[3:18] go with Clippy's Scrumptious Cookie. Down below, 
let's click on add. It's now ingesting all these  
[3:23] different sources into its knowledge base. Next, 
we need to define what should happen if it can't  
[3:29] find a response in our knowledge source. Now 
by default, it says to generate an AI response  
[3:34] without knowledge sources. Now, I'd rather have 
it show a custom message, so right here, I'll  
[3:38] select show a custom message, and it'll simply 
say, "Sorry, I don't know." Again, I don't want  
[3:43] it making up information. In the bottom right-hand 
corner, let's click on save. There we go, we now  
[3:48] have our first source. Let's try adding one more. 
In the top right-hand corner, I'll click on add  
[3:53] source. This opens up that same dialogue again. 
Let's go back to the type dropdown, and this time,  
[3:58] I'd like to add an FAQ file that I have stored in 
Google Docs. I'll click on this option, and you'll  
[4:04] have to start by connecting your Google Docs 
account. So, over here, let's click on connect. In  
[4:09] the top right-hand corner, I'll add a connection. 
Then select your Google account. Once you finish  
[4:14] setting up your account, you'll see the connection 
right here. Down below, you can now select the  
[4:19] file that you would like to include as part of 
the knowledge base. I’ll click on this, and here,  
[4:23] I see the Kevin Cookie Company FAQ. I'll select 
this file right here, and in the bottom right-hand  
[4:27] corner, let's now add the source. Right up 
above, you can see that we now have two different  
[4:32] sources. Our bot's getting a lot smarter. Before 
we move on, let's look at some of the settings.  
[4:37] Right up here, we have knowledge settings, and 
we looked at this earlier. This is the fallback  
[4:42] response when it can't find an answer. I'm good 
with this, so over here, let's cancel. To the  
[4:47] right of that, we also have schedule all. This 
allows you to schedule how often Zapier refreshes  
[4:53] these knowledge sources. This is useful if you 
have content that changes over time, like let's  
[4:58] say a menu or maybe new policies. You can define 
how often it checks, the day that it should run  
[5:04] on, and right over here, you could also specify 
the time window. Once this is set, the bot will  
[5:09] automatically stay up to date without me having 
to manually upload anything. Right over here,  
[5:14] I'll click on save, and the beautiful thing is 
the knowledge will stay fresh all on its own. So,  
[5:18] if I update the Google Doc, that information will 
automatically flow to my bot on the schedule that  
[5:24] I specified. Let's now shift to the logic tab. 
Over here, I'll click on logic. This is where  
[5:29] we can add extra actions and controls to your 
chatbot. Right up at the very top, we can collect  
[5:35] contact information from users. Underneath that, 
we can add buttons that trigger a Zap. Here we  
[5:41] have a Zap, and at the very bottom, we even have 
suggestions that help guide the conversation.  
[5:46] If you haven't heard of a Zap before, maybe you're 
wondering what that is, it's an automated workflow  
[5:51] in Zapier that connects two or more apps together. 
If you want a full walkthrough on how Zaps work  
[5:57] and how to build them, I have a dedicated Zapier 
video linked right down below in the description. 
[6:02] In my case, if the chatbot can't answer a 
customer's question, I want to make sure I  
[6:07] collect their information so I can follow up. So, 
right up on top, let's click on collect leads. On  
[6:13] this page, I'll create a new table to store 
this information. Down below, I can see all  
[6:17] the different types of fields that the chatbot 
will collect if it doesn't have an answer for  
[6:22] the customer. So, I have the name, the email, the 
phone, and the company. Now for a cookie business,  
[6:25] I don't know if I need a customer's company, so 
over here, I could delete that field. If we click  
[6:30] on add field, I can also add additional custom 
fields. However, I don't need any others. I think  
[6:35] these three work well. Right here, I could define 
when to ask for this information at the beginning  
[6:40] of a conversation, after a few messages, when 
certain keywords are used, but I really just  
[6:44] need this when the chatbot cannot provide an 
answer, so I’ll select this option. Down below,  
[6:49] everything else looks good, so let's click on 
create logic. This opens up a new tab where we  
[6:54] can see the table with the three fields, the name, 
the email, and the phone. This all looks good,  
[6:58] so let's jump back to the chatbot. Right up 
above, let's go back to the previous tab. The  
[7:03] logic looks good, so over on the left-hand 
side, let's now move on to the theme. This  
[7:07] is where we can adjust the look and feel of the 
chatbot, so you could set things like the avatar,  
[7:12] the logo, the embed icon, and down below, you 
could also adjust the colors. Right up on top,  
[7:17] let's choose an avatar. Click on choose image, and 
here we have the chip avatar. Let’s click on open,  
[7:23] and that'll now upload him. Let's go down a little 
bit more and I also want to customize the color.  
[7:28] I'll click on custom color, and let's go with 
white. I think this all looks good, so down below,  
[7:34] I'll click on save changes. Over on the right-hand 
side, we can preview what it looks like. Here we  
[7:38] have that nice modern and sleek white background, 
and we also see Chip the avatar. Look at that,  
[7:44] he's waving at us. How cute. Over on the left-hand 
side, let's now shift into integrations. This is  
[7:49] where you can connect your chatbot to services 
like Slack, Facebook Messenger, Gmail, Zendesk,  
[7:54] or you could even build your own from scratch. 
These options let the bot send or receive messages  
[8:00] through these different platforms. Now you could 
even build your own integration from scratch  
[8:05] if you want even more advanced workflows. 
For this video, we're going to keep things  
[8:09] simple and just focus on the chatbot itself.
So, I'll skip all these integrations for now.  
[8:14] Over on the left-hand side, let's now shift 
into the AI model. Here you can choose what  
[8:18] model powers your chatbot. Right up on top, we 
have two different options. You have OpenAI,  
[8:23] GPT 4.1 Mini, or 4.0 Mini. Those are the default 
options. However, if you would like access  
[8:29] to even more models, right up here, you could 
connect your OpenAI or your Anthropic account,  
[8:35] and then those different models can power 
your chatbot. Down below, you can also set  
[8:39] the creativity that your chatbot can take. Now I'm 
going to stick with all the defaults, and this all  
[8:44] looks good. Now that everything's set up, let's 
try out the chatbot. Over on the right-hand side,  
[8:49] I have the preview window. Here I could talk to it 
just like a customer would. Let's start with a few  
[8:55] common questions to make sure it responds the way 
I expect it to. Let's ask something simple. Down  
[9:01] below, I'll type in what are your store hours, 
and then let's send that through. Right up above,  
[9:05] I can see the store hours, and everything appears 
to be accurate. This lets me check that it's  
[9:09] pulling from the knowledge I added earlier. 
Now let's try something it shouldn't know. So  
[9:15] down below, let's type in another question. How 
many employees work at the Kevin Cookie Company?  
[9:20] We don't have that information on the website, 
and it's certainly not in the FAQ. Let's submit  
[9:24] that. Since it doesn't know the answer, right 
up on top, it says, "Sorry, I don't know." And  
[9:28] then it also asks me for information, like the 
name, email, and phone number so someone on our  
[9:33] support team can follow up. Testing here is quick 
and easy. Now that the chatbot's ready, let's try  
[9:38] adding it to a website. In the top right-hand 
corner, let's click on the button that says,  
[9:43] "Share," and we have a few different options. 
You can invite other people to work with you on  
[9:49] this chatbot. Now, I don't need any assistance, 
so over here, let's click on Public Link. Here  
[9:54] you can get a public link that you can share with 
others, and that'll bring them to Zapier, where  
[9:59] they can then interact directly with the chatbot.
Right up on top, we can see what this experience  
[10:04] looks like. And if we go back, you can also 
embed it directly on your website, and you have  
[10:09] two different options. You can place it on your 
website as a pop-up, and here we see that option,  
[10:15] or you can place it in line onto your website. 
Now I'd like to place it in line on the site,  
[10:20] so I'll click on this option, and down below you 
can get a script, or you can copy an iframe. I'll  
[10:25] go with a script right here, and let's copy the 
embed code. Now go to your website builder. I'm  
[10:31] using Google Sites, but this works with WordPress, 
Wix, Squarespace, and many others. Now right over  
[10:37] on the right-hand side in the Insert menu, I'll 
insert embed code, and let's go with embed code,  
[10:42] and here I can paste in the code, click on Next, 
and let's insert that onto the page. Over here,  
[10:47] let's center it. So, it's right here in 
the middle of the page under Support,  
[10:50] and there I can see the chat bot appear on the 
page. In the top right-hand corner, I'll now  
[10:55] click on Publish, and let's see what it looks 
like live on the site. I'm now on the website,  
[10:59] and here I see Chip the Cookie Helper. Let's jump 
down to the bottom, and let's ask a question.  
[11:04] Where are you located? And then let's submit 
that. Perfect. Look at that. 123 Sweet Street,  
[11:09] Cookieville, USA. Back within Zapier, I can 
view all conversations. This is where I can  
[11:15] see how customers and people are interacting 
with the chat bot that we just built. Over on  
[11:20] the left-hand side under the Review category, 
let's click on Conversations. Here I can check  
[11:25] what customers are asking. We can check for any 
issues, and we can also make improvements based  
[11:29] on those interactions. For example, here I asked 
where the store is located, and the bot responded  
[11:35] with the correct address from the knowledge base. 
Once you start having more and more conversations,  
[11:39] you can also check how it's performing over time. 
On the left-hand side, let's click into Analytics.  
[11:45] Here you can track the number of conversations, 
messages, the number of links that were clicked,  
[11:50] and as more people use the bot, this page will 
fill in with trends. Here we have top keywords,  
[11:55] the sentiment score, and you have additional 
data. This helps you understand what customers  
[12:00] are asking and how well the bot's responding. 
And just like that, we now have a fully working  
[12:06] customer support chat bot for the Kevin Cookie 
Company. That was really easy. If you want  
[12:11] to learn more about building Zaps or connecting 
this chat bot to other apps, I have a full Zapier  
[12:16] tutorial linked in the description. Thanks for 
watching, and I hope to see you in the next one.