sYLik9uuP7I
Let's build with Zapier and AI (100+ videos)
Transcript
[00:00] Are AI agents better than MCPs? Let's find out in today's video why we would leverage either one, their use case, and how to do it automatically. Pretty cool stuff. Let's jump into today's video. Best answer this question. We're going to be checking out Zapier AI agents and Zapier's ability to create MCP servers that allow us to do a bunch of cool actions across the interwebs for over 8,000 apps. Today's video is sponsored by Zapier because, as you already know, we have teamed up and we are doing a bunch of cool stuff together. MCP or AI agent, which do we like? as a high level of the difference between agents and MCP
[00:30] servers is MCP servers typically require manual input. Eg. You actually got to be there on your laptop, on your phone, on your desktop, type something in to get information out. Agents, on the other hand, and what I'm going to show you in this video is we can set up AI agents to do stuff automatically on our behalf, which is pretty cool. You're going to see a pretty specific example that I will use personally when it comes to travel. We use MCP servers in our workflow in order to expediate productivity when it comes to task. Now, I'm going to leave a video in the description down below that shows you an
[01:01] MCP server that we created together for vibe coding and getting better outputs when it comes to API and connecting and creating really cool apps. This little MCP server that we created with Zapier here gets the most up-to-date information when it comes to code. Alternatively, you may have used MTCP servers in cool things like claw desktop, giving us the ability to get a bunch of data on a topic and then maybe push it to a Google sheet with Zapier. So, these are cool, but let me show you what makes AI agent stand out when it comes to this kind of workflow. As one thing, as you know, is it still requires
[01:31] me to say hello and that rhymed on accident, but hello, I'm typing I'm here. So, let's come over to Zapier AI agents and see what we can do here. Essentially, what we're doing is we're making the agent ourselves or we're making a duplicate of ourselves that does the work for us and you could be cooking a really good steak or honestly preferably a brisket on the grill. You already know how that goes. Salt, butter, pepper. We like it. Our agent today I'm going to create is going to be travel agent. So, one thing about me which you may or may not know is I like to travel and I like to go to different
[02:02] places but on a budget, right? I want to make sure I get the best deals and the best places, etc., etc. So, let's create a action that's going to run automatically for me in the background. And as a quick overview, look at this. We can visit websites, Google search, and add very specific actions on over 7,000 apps. That's a lot of apps. We're going to come up here to configure and we're going to create a behavior. So, the automation I'm going to create today with these AI agents would be I would go to something like Claude and be like, is there any good deals that just came out for travel in any place in the world? But that would require me to basically
[02:34] establish that in my workflow every single day. Hey, I wake up in the morning, you know, get you a good cup of coffee and proceed. But let's just let AI handle the heavy lifting for me. So, my trigger here is I'm going to follow a very specific data source that provides this information. So, we're going to do an RSS feed. RSS by Zepier. We're going to say new item and feed. If you don't know what an RSS feed is, I want you to think of it like when a new newspaper comes out. Like, you know, you're walking down the street and it's like new newspaper and it's like the date that it came out is probably today. Same thing is an RSS feed but in a digital
[03:05] way. Whether it's a new blog, new piece of information. This is how we get data in a automatic software way comparative to like you go to your news stand. Now you got the most up-to-date newspaper. That make sense y'all? Or am I going crazy with these analogies? Let me know in the comments down below. Now you may want to use this in different context, but for me, I just typed in top RSS feeds for deals. I'm going to go to this top blog here, and I'm simply going to click this. Now, I know this looks a little crazy, but just come up here and copy the link. Going back over here, I'm gonna put in this feed URL. I'm going hit save. So, now our AI agent's going
[03:37] to run every single time a new item is in this feed. And one thing you should know about AI agents is obviously requires no coding. It's simply us just talking to it. So, you could go as far as basically saying, "Give me an example of an output from this RSS feed." I'm going to hit save here. And right behind me, we're going to test this behavior. And this will walk through step by step exactly what it's doing. And here we go. Here is an example of the output from the RSSV provided. Up to 50% off NYC attractions and shows. And we get a nice little link here. And the publish date is right on time. It is May 14th. An
[04:08] experience all New York has to offer special savings on shows, spas, restaurants, and activities in the city. Pretty nice. Let's say we want to set this up so that every single time we receive some news like this, we do a couple different actions. Here's the idea. And here are the steps we're about to outline. It would have been the same exact steps we would have used in claw desktop in the MCP server, but now we can just let the AI agent do it for us. So, here are the actionable steps I would had have done manually, but now we let our agent do it. I want you to search online using Google search. And to find this specific tool, just come up here to insert tools. Find the tickets for flights from Dallas DFW to whatever
[04:40] location is mentioned in the RSS data. Output top 10 things or output 10 top things I should do in this location that is mentioned in the RSS data. Notice that I'm referencing RSS data because it's know it's going to be from this trigger here. And with this information, send it here. So, I'm going to put in my email. And we can do that with add tools. Gmail. Send email. I'm going to go ahead and choose my account here. Subject can be let AI generate the field. Let AI generate the body. Nice. Save. Actually, one thing we got to make sure it's sending to the right person. So, we are going to send this to
[05:10] ourselves or alternatively send this to whoever you want. Made it save. Save again. And let's see if this will work. Retest behavior. So, we just got something for RSS feed, new deal for travel. It's doing Google to give us a Google search here to find flight tickets for this. I didn't specify a lot here, so I'm really curious on the dates it's going to choose and how to proceed from here. We'll see what that comes out to in the email. So, now it's searching Google for the top 10 things to do in New York City. And it should send us email here. Sending email. When we're in test mode, it's going to prompt us to confirm whether or not to do an action like this. But when it's live, I'm going
[05:41] to hit approve. When it's live, it'll do this automatically. So, coming over to Gmail here, we got flight tickets and top things to do in NYC. Nice. And here we go. Wow, this is actually really in-depth. So, not only are we given links to the underlying tickets, which is cool. So, I can literally click it and get the exact link for it. Also want to point out that if I would have switched it from plain text to HTML, this probably would have been formatted a little bit better. So, that's what you can do in the action of Gmail as well. Eg. Come in here, go to settings, and if you come to body type here, we can set this to a specific field. Choosing HTML
[06:12] here will format the body of the email a little bit better, but for now we still got our email. Then we got the top 10 things to do in NYC. Pretty nice. And there we go. Best regrets, Corin Brown. I guess I sent it to myself. So now that it works, we can set that to be live and it's going to work around the clock. And that's one of the advantages of using AI agents over MCP servers as it requires no manual work. In addition to that, we could have sent this from another email. We could have sent this to a Slack channel. We could have had it texted to us. Whatever it may be, those are two random videos. That is my face. Make
[06:42] sure to leave a like if you felt like you learned something in today's video.