Zapier AI Tutorial for Beginners (ChatGPT, Claude, Gemini)
Let's build with Zapier and AI (100+ videos)📅 2025-03-15
Transcript
[00:00] In this fullstep guide, I'm going to show you how to use every single tool within Zapier and more specifically how to leverage artificial intelligence within Zapier. When it comes to your choice in the type of artificial intelligence you use, don't worry. Zapier integrates with them all. Whether that is OpenAI, Google Gemini, or Claude, let's go and jump in. Welcome back y'all. In today's video, we're going to do a full-blown in-depth tutorial on everything you should know about Zapier and implementing it into your business, especially in the context of AI. As we know, we need to understand how to prompt correctly, the type of automations that currently exist with
[00:30] businesses when it comes to AI automation and everything you should know when it comes to cost when adding this extra layer to your business. As you already know, me and Zapier have teamed up and they're like, Corbin, we got to do a bunch of cool videos together. I'm like, yes, I agree. So, today's video is sponsored by Zapier. We're going to do a full-blown in-depth tutorial just for you. So, let's go ahead and jump in. I want to make sure I leave a link in the description down below so you can go ahead and sign up to Zapier for free. What's great is that they have a free tier so you can just kind of play around and see if you actually like the platform. And this is a long overdue video because the last time I did a video on this topic was two years ago. But as you can see, Zapier
[01:01] has expanded drastically and their ecosystem is getting very very big. So we got Zaps, tables, interfaces, chatbots, canvas, and agents. But step one here is Corbin, how do we even integrate AI into this platform and leverage it? To do that is very simple. We're going to come over here and select app connections. Now, what you'll notice is that right now we can search up Claude. There we go. Gemini. Boom. And if you want an indexed version of this of all the AI that Zapier currently integrates with, I'll make sure I leave this in the description down below as well. This is Zapier's app catalog, but obviously as you can see to the left of
[01:32] me, there is a ton of other apps that we integrate with when it comes to Zapier. This is just cool to look through, check out, and just understand the capabilities of this platform. And there we go. The one we'll be using today is going to be Chat GBT by OpenAI. So, coming over here to apps, I'm going to do Chat GBT. Here we go. And you were like, Corbin, does AI really work? Well, look at my connections. 22 zaps, 167 zaps, 82 zaps. And to take a step back, a zap in this context is just one automation. For example, one flow would be like an automatic AI email responder, which I've done a video on this channel about, but for now, let's add this
[02:03] connection. This is going to require an API key, which gives us the ability to access the AI from OpenAI. I might be saying AI a lot in this video, so get used to it. It's going like AI AIO. I'll go ahead and leave the link to the OpenAI developer portal in the description down below. This is where we're going to create our API key. So, I'm going to hit sign up here. or for me, I'm actually going to hit login, but you go ahead and sign up. It is free to use just to sign up an account. The steps and processes I'm going to show here work with Enthropic and Gemini as well if you want to use that as your AI model for these workflows. Once you're logged in, go to settings and come to
[02:33] API keys. We're going to hit create new secret key. For the key name, I'm just going to put zap your key. This is for your reference. Select your project. For me, it's just default project. And very important for permissions, make sure it is all so we can do stuff with it. With that all selected, hit create secret key. You'll then get a pop-up showing your key. Just hit copy. Come over here and hit paste. Then hit yes, continue with chatbt open AAI. I got my new connection here. To rename it, I'm going to hit edit connection name. And as you can see right there, we named it zap zap zap zap. If you ever face a situation where this key doesn't work, maybe you forgot your key. You could have pasted
[03:03] over. Just go through that workflow again. Recreate the key. Add it here. But that's it y'all. Now we have the ability to do a ton of cool stuff. Now let's actually use it. Let's check out zaps. What should we know about using AI within an automation? I'm going to go ahead and check out a zap I actually did earlier on my channel here, which was for email AI design. And you might be asking yourself, Corbin, what do you mean AI email design? I showed you how to use artificial intelligence to create HTML code if you have no coding experience and get a really awesome design for your email. Now, I use this personally within my company and I think this is really powerful as end of the
[03:34] day, we always don't like plain text emails. And let me just show you what it looks like. Retest that. With the power of AI, we were able to get a really awesome header here. Nice UI button. You could actually click an external link there. Everything like that. But when it comes to adding AI in this workflow, so let's say for example, we don't want everything to be fixed text. Eg this email, we want to personalize it depending on the relevant candidate that we're sending it to. The example automation I have here is very specific to onboarding a new employee. Therefore, let's do that. How Zapier works is that every single automation is going to
[04:06] start with a trigger and then everything subsequent to that will be what we call actions. A trigger is an event that happens on an external software that we've integrated with Zapier. like you saw with the Chad GBT, but this could be Google Workspace admin, Gmail, Instagram, Facebook, Google Sheet, the list goes on. But the idea here is that when something happens, everything subsequent will incur. So for this example, when we create a new user in Google Workspace admin, boom, this is going. And that's the beauty of software. Everything that we have
[04:37] subsequent to this trigger will happen automatically. I could be asleep. I could be cooking some really good ribs on the grill. You already know how that goes. You want that spicy barbecue sauce, lather it up. You want a really good corn recipe? I got a really good one. Texas steak seasoning, butter, heat it up in the microwave. Go to that grill. Lather up. Now you got some good corn. Texas corn. From here though, let's add Chad GBT. Obviously, if you're using Clawude, you would do Claw like this. We're going to do Chad GBT,
[05:07] though. Our first step here is selecting what the heck it's going to do. Like, what do you want Chad GBT to do? With Zapier and the power of Zapier, we can do this all of no code and we can do a ton of stuff. I mean, this is way more than it was two years ago. And that's part of the reason why I wanted to make this video as we have the ability, for example, to create an assistant. We have the ability to create a transcript. We have the ability to conversate with an assistant. And you're going to be like, Corbin, take a step back. What the heck is an assistant? Assistant is what we call assistance API within OpenAI. These are specialized models trained on your data to give better and better answers.
[05:39] If you're like, I want a video on that, just type in Corbin assistance API. But the point is with Zabier, we can basically functionally do everything we would ever want to do with AI with an integration workflow. What we're going to do today is conversation. Select your account. Zap, hit continue. That workflow of selecting your account and the action will incur with every action block you use outside of just the AI block. Right? This is kind of the workflow within Zapier. It will identify the specific action it's about to do. So for us, we know it's conversation. And from here, this is where we do our prompting. This is prompt engineering. Okay? This is what you've always heard
[06:09] about on the little interwebs where it's talking about prompt engineer makes thousands of dollars. This is prompt engineering. What I'm going to do is I'm going to show you proper prompt engineering techniques. The first thing we like to do is context. I want you to think of context as if this AI is a human. And we need to tell this human what they do and what they ought to do. So the context for this assistant is that you help onboarding at bumpups and write emails. If you're like Corin, what the heck is a bumpup? That is a software company. You're going to check it out. Point is once we give context this could be anything. You are a customer service
[06:40] rep at XYZ company. You are a social media marketing manager that is specialized in writing captions. What's the context? What do you want to do? The next part here is going to be steps. So we're going to do instructions here and we're do step one, step two, step three to however many you want. And the idea here is in theory what I would typically do is kind of load this with some variable data in a previous block. For example, let's go and do that. The idea is maybe we want to load a Google doc of context of what bumpups is and more information for it to write a more
[07:11] personalized email for the new employee. So the reason we'll do a Google doc here is because of the fact that this could be data that changes over time rather than a PDF that's static and you have to reupload the PDF. It's very easy just to come to a Google doc, change the data and you're good to go. Works right in the workflow. So I'll just be like bump ups context. And in reality because we're onboarding a new employee, we need to give context of like what the company is, what it does, etc. So go and grab that wherever that would be located. For me, I'll grab it from our LinkedIn here of bumpups. So we got a nice little paragraph here and obviously this can be a lot longer. I've shown you workflows where you could do FAQs and reference
[07:43] that as the data because frequently asked questions in customer service situations are very helpful for the AI as we don't want to give bad bad information. But for now, let's proceed. Typically, what I like to do in my videos as well is I like to gut check to show to you and to prove to you that it's actually using this as data. Sometimes we feel like the AI can hallucinate and make up stuff. So therefore, to confirm that, we're going to make sure that we have the specific detail that this company was founded in 2023 Dallas, Texas is identified within the output. First off, we need to provide the data. Like if anything, just
[08:14] integrate your Google Doc. That'll be good to go. We're going to configure here and we're going to find that specific document name. For me, it's bumpups context. Continue and let's see if it finds it. There we go. It was successfully found. But your next question might be, Corbin, where the heck is the text? What is all this jargon? Don't worry, we'll figure it out. We're going to add another block here. Now, what's nice about Zapier is they actually created a pretty new block here, a couple months old, called files by Zapier. This allows us to extract data from PDFs, which I've done in a video, but also Google Docs. So, we're going to do an action event of text from file. Continue. The file itself is going
[08:45] to come from the previous step here. So, we're going to say find Google Doc, and it is file text exist, but not shown. Continue. Test step. And boom, we have our paragraph here. What's super nice is that Zapier took the initiative to make the text from file. Originally, we had to do this of code, but now it's a simple drag and drop. And honestly, since we're on the topic of text from file, and you might be saying that's not an app, Corbin, that's not a third party app. That is a Zapier app. And what I mean by that is Zapier has its own little toolbox. If I go to app families here, we can scroll down and click Zapier. This right here is another fundamental part of the Zapier
[09:15] automation platform. As this is your toolbox, very, very powerful stuff here. But you can do stuff like filter in a zap, format in a zap, schedule a zap. Learning this is kind of laying the brick layer of really giving you the ability to do crazy stuff like I just did where you're just like, "Wait, I got a Google doc. Extract text from doc using a zap year family block." Did that rhyme? Maybe. I think that did. Let's keep going here though. Once we have that, we can now add this as data within our prompt here. When AI was first being integrated two years ago, what I'm about
[09:46] to show you wouldn't have worked because of the fact that the what they call what the engineers call context window uh was too small. We weren't able to put that much data into a user message. Therefore, you know, kind of put us in a box. Now, you can really frontload this. So, I'm going to say this here is the context of bumpups parenthesis and then we are going to input the data from the previous block. So, we're going to add here and we're say text from file text. So what you're noticing and why automations are powerful is this is very much dynamic in the sense of anytime we
[10:17] add data from a previous block that could change depending on the circumstances for this context it is fixed but in theory if I went to my bump context here I added a couple more sentences don't worry the next time the automation flow is going to work and they grab the data from that doc again this will be updated and that's powerful think of the instructions like that earlier example here where we are going to treat a zap year automation like an employee. Idea being that if you were to train an employee how to respond to a customer service email, what is the SOP, standard operating procedure on how to
[10:48] do so? In this one, I just kind of want to make like a nice little friendly hello paragraph that's a little bit personalized based off the context of bumpups. This is how we're going to structure up to this point. Context, you're helping onboarding at bumpups and write emails. Here's the context of bumpups. Dynamic field instruction. Step one, learn what Bumpups is and when it was founded. Step two, write a short intro paragraph email for a new hireer at bumpups explaining this. Oh, we got to capitalize bumpups. Uh oh, there we go. Step three, output just a paragraph, nothing else. Step three is fundamentally very important for you to put in a prompt as what you'll notice is
[11:19] that when you are testing and playing around, sometimes these AI models can say too much. What I mean by that is let's say we're making an AI automation workflow for 10 social media captions and you know we get to the point of an output. Sometimes the model will give you the 10 social media captions, but at the beginning of the output slash the end of the output, it'll be like, hey, here is your 10 social media captions. But we don't want it to say that. Don't say that. Just give me the 10 social media captions. This is actually very important at scale as well to ensure consistent outputs. Now, the last part here is we're going to add a formatter block. Formatter block is very specific
[11:49] ways that we want to make sure is consistent in the output. Now, an example of this is that this intro paragraph, not intro paragraph, intro, I want a max of like three sentences. I don't want this thing to ramble. Don't talk too much. Just give me three sentences. So we do max of three sentences for intro paragraph. I also want you to reference the idea that when I identified that I want three sentences, I also identify a previous piece of text in this prompt. Intro paragraph. The AI will have enough context to realize I'm referencing the same paragraph here and here. This isn't two separate things. Now it comes into
[12:20] the model. And your choice and model is up to your discretion, but here is a quick rundown. Every AI company will typically have two major types of models. Obviously, I I know I know there's tons of models that exist, but there's going to be two major types. There's going to be the cost-effective one that's very fast, give me an output. There's going to be a higher level model that's going to be more expensive and takes more time. It's more reasonbased. When looking at this, we have more discretion on which one we choose. But in reality, I would probably opt for something like GBT40 Mini as it's fast. And where we've gone when it comes to AI
[12:52] is tremendously improved than what it was 2 years ago. If I was making this video 2 years ago, I would have told you to do GBT4 and not GBT 3.5 as 3.5 outputs weren't effective at creating humanlike writing. But now we're getting to the point where AI is just getting just really good, right? So we're going to do GBT4 mini. If you wanted to app for a higher level model, you're going to be looking at the 01 series by chat GBT here. here. So, 01 mini, 01, etc. That's for very, very in-depth work and
[13:23] in-depth outputs. And here's the best part. If you don't even know which one to choose, just plug it in here, switch it, see the output. I like this one. I don't like this one. Oh, I mean, if you want something really crazy, come down here, click that GBT 3.5 Turbo, and see what I used to have to deal with. Okay, two two years ago was no fun. I mean, it was fun. Okay, it was definitely fun, but the outputs were nowhere near what they are now. Next, most important part is going to be a member key. This is going to be a random string of 32 characters like this right here. Right? Uh I'm going to just say email. Boom. And the idea with this is this is going
[13:54] to ensure consistent output. So if you get an output you like, think of the memory key as what's locking it in. Like you don't have to worry that 10 outputs down the road it starts going crazy on you. So we get a good output here. We're locked in with that memory key. Past this. Everything is pretty good. So I'm going hit continue here. We're going to test this out. So scrolling down here. Let's see our output. And boom. Welcome to Bumbups. We are an innovative company founded in 2023 in Dallas, Texas. Remember I pointed that out. This proves that it actually is reading the document. Surprise, right? 2023 Dallas,
[14:24] Texas. Amazing. That harnesses the power of AI to transform video content into dynamic interactive experiences. Our mission is to make a complex data more accessible and actionable, allowing users to engage with videos in an unprecedented way. So, what I can do now is with this AI output, I can come to this email that's for an intro for a new employee. I can go to configure. I can scroll down here and I can input it as one of these parts here. So, for example, we're thrilled to have you at XYZ company. Like, no, no, no, I don't want that. I can now add it as a dynamic field here. Computer plus conversation. And what is important for you to
[14:55] understand is that the one you're looking for is reply. Boom. And here we go. Let's test it. And it shows up in our email here. As a side note, because we're on the topic of AI bumpups, if your business has a workflow that requires you to analyze a video, API is going to be released pretty soon here. and we're going to be able to integrate it into Zapier as well. Plan on doing some videos on that. Therefore, don't worry. You don't have to watch that video anymore or that meeting or anything. We will analyze it and give you what you want. This covers on a service level what we need to understand about automations. As you could probably already tell, I have a 100 plus video
[15:26] playlist that goes over a ton of stuff with Zapier and specifically of AI. So, check that out. Let's do a quick dive of the other features found within Zapier. Tables. I want you to think of tables as an upgraded version of Google Sheet. And if you're a developer, think of it like Firebase, Fire Store. Like this is a data oriented infrastructure built within Zapier. And here is the advantage. Right off the bat, as you can see right up there, you can actually link automations to this database or alternatively tables. But end of the day, it's a database, right? There's data, maybe a base, who knows? Idea
[15:58] though is that we can set up columns here for very specific variables. And you know what's really cool and what I've shown in videos is our ability to set up an automation within the table itself. Now, the idea here is that let's say we have a form here that's taking in data. So, you know, cool, nice phone number 1111, company size 1 to 10, and additional notes. We're like, okay, this is cool. When a bunch of data comes in, a bunch of records, I have the discretion on whether or not when I click this boolean checkbox to send an automation that sends them a follow-up
[16:29] email like like, hey, I saw you fill out the form. This is cool. Let's do it. The idea being though is if I get a record that I don't want to send the next step to XYZ process, I don't have to check that bullying there and it won't do the automation. So this gives more discretion in that way with every feature about I see right now. What's super cool about Zapier is it's free to use. You can just try it out, see if you like it and proceed. That covers tables. That is your short quick one minute on tables. Let's go to interfaces. I want to point out as well with tables everything you saw with that artificial intelligence logic in the automation
[17:01] will obviously apply to those zaps that are integrated within the table itself even have the ability within actual cells to leverage AI in any way. Now coming over to interfaces here I like this and the reason I like this is that think of this right now as your ability to just create really really cool stuff either internally into your company that integrates within Zapu's ecosystem or externally for your company for a landing page. What's an example of this Corbin? Well, right off the bat, we have a nice little drag and drop builder here that we're able to make forms that are connected to Zapier, Cabons, media, you know, very, very standard stuff within a
[17:32] landing page that you'd want to show up. But here's a real example of how we actually use this within bumpups. As it stands, I use interfaces within bumpups, specifically for the support tickets and contact. Why do I do this? Well, one, you can just make really nice UI landing pages very fast, very easy. But two, which is really cool, is that when a user fills out this form, it's connected to my table records and is connected to my automations associated with those table records. Really, really cool way to start automating your customer service. Now, we get to the chat bots. Chatbot. Now, if you know me, I did a
[18:04] crazy deep dive when these things first released all about this. I I love these things. These things were awesome. We were able to leverage these chat bots on landing pages and just create really cool stuff. Like I created Grammar Bot Brew, financial bot brew, ad campaign bot brew. The idea behind the chatbot here is that one of course it is integrated within the ecosystem of Zapier. So we get access to everything I just shown up to this point. But second this is a chatbot we can use and leverage within our website or on a landing page. We have the very powerful stuff of our ability to give context to the chatbot whether that is a web page
[18:36] upload file or of course those tables all that data in the records and then the actual functional abilities that we can do outside of just like text in like the customer says hello hello what do you want but actually has the ability to do uh different actions here as well collect leads zap buttons run zap and then as we always love with any type of software UI or ability to add branding next we get to canvas up to this point I've shown you a lot in Zappy's ecosystem Don't feel overwhelmed though as I would just suggest you to start on ground zero and just like let's build a zap year
[19:06] automation. Build a zap solve a specific issue within your business and grow from there because that leads to canvas. And what canvas is is really just connecting it all together. Right? We can leverage canvas to build out how we want the workflows to look for a said painpoint for our company. Whether we're using interfaces for the workflow and then after interfaces and a form field then we'll use a zap. Oh, then we'll take that data and put it into tables. What this right here does for us and obviously the if else logic here, which is kind of cool. What this is is like
[19:37] our custom little whiteboard for our company to kind of draw out what we need to do. So, for example, hey, we're getting a lot of emails. We need to solve this Gmail automation. What is going to be the workflow of how we want to approach this? Brainstorming. This is cool. This is free to use. And then this leads us to agents. What is an agent? No, not James Bond. Not 007. Now, I said prompt engineering was in your feed, but maybe not as much as it used to be. But I know for sure you are seeing AI agents everywhere, everywhere, everywhere. AI agent here, AI agent there. But what I
[20:08] can tell you about these AI agents is these give you the ability to actually integrate and access over 7,000 apps. Every other AI agent that you've seen up to this point seems to be siloed. It's like, oh, it's just doing text input, text output. No, no. These I was able to create a stock trading bot, which I have a video on you can check out. crypto trading bot. I was able to create a bot that scraped competitors websites. I was able to create a bot that scraped a potential client's website, got all the relevant data, automatically created out of Gmail and was able to send it to said client because it was scraped the data
[20:38] of the email. Crazy stuff. This is the future. This is super cool. We are I'm able to do I did all this for free. So, you're able to play around with this and understand this at a conceptual level and just get going. These agents are extremely powerful. Now we have the ability to not only get a cool little, "Hey, the AI agent can talk to me." No, this thing can do task across the internet, which we haven't really seen before. There is your beginner guide to Zapier. A lot has changed in the last 2 years since the last time I made a video on this topic. Their ecosystem is
[21:09] growing and you can see a huge push towards really automating a lot of workflows you can find within a business. And here's my last point on the topic of choosing an automation software for your business. and why Zapier kind of stands out to me and they have been for the last couple years. If you're going to take the initiative and start building out workflows that automate task and labor within your company, you are opted and you probably are more inclined to go with the company that is going to give you the most capabilities long, mid, and short term. What I mean by that, a lot of
[21:40] competitors when it comes to automation workflows, they just do the automation and even then they don't integrate to as many platforms as Zeper does. The big play here is that ecosystem. That right there is how you build fullblown workflows that you otherwise couldn't with other softwares. Regardless though, make sure to leave a like if you found value in today's video. I'll leave all the relevant information that I said in today's video in the description down below. All the links. I'll see you in the next video. Zapier AI automation.
[22:11] Two random videos. That's my face. I'll see you in the next