Create Your AI Social Media Agent in MINUTES! Zapier Agent Tutorial - Build AI Employees w/No Code

Today I'm going to show you how you can make your very own AI social media manager. That's right, over the next 10-15 minutes or so, you're going to learn how you can build your very first AI employee, and you don't need any technical skills at all. This employee is going to go out and do research about your niche, look at the latest news, and then it's going to write high-converting, unique Facebook, LinkedIn, as well as Instagram posts for you, all on automation. It's just so mind-blowing how quickly this technology is advancing on basically daily basis. Also, if you're new to my channel, hi, my name is Helena. And on this channel, we talk about everything AI automation. So please make sure you like and subscribe below to keep up with the latest updates in AI. Alright, so without further ado, let's dive into it. So to build our AI agent, we're going to use Zapier Agent Platform. And I really like this platform because it is the easiest one to onboard and you don't need to write any code at all. So if you already have an existing Zapier account, this is actually absolutely free for you. Zapier the OG in the automation industry. So you know that it is the most stable platform out there. So go to agent.zapier.com if you do need a new account, and also going to include the link in the descriptions below as well. And here we can just start a new agent. And you can already see that there's a number of prebuilt templates here. Now in our example here, we're going to start a new agent, we're going to call this social media agent. And we're going to start from scratch. Now by default, after you create your agent, you're going to land on the Chat tab. We're going to go over to the Configure tab though because we need to tell this agent what we want it to do. We are going to add in all our data sources and behaviors, and that's it. So this is a super simple way to get started with agents. In the data sources, this includes all of the things that you want this agent to know, and you can link it to all of these various different data sources from Airtable to Excel to Google Docs to HubSpot, Notion, Zendesk, etc. This is really really great because in the past you would have to export all of that information in a text file or a PDF file and upload it into your agent. And whenever the information changes, you need to then redo the whole process again. But with Zapier Agent, you can link it right in with your Notion account or your Google Docs. So when that information changes on those documentations, your agent's knowledge base will change as well. It is completely optional to link your agent to a data source, so you can have agents without a data source. But for our example here, we're going to link it to a Google Doc. This is what my Google Doc looks like. It has several examples of high-converting posts. So I'm going to teach this agent about how to write these high-converting posts. To link my data source to my Google Doc, I'm just going to click on Google Doc. If this is your first time connecting your Zapier account to a Google Doc, then you need to authenticate by putting in your username and password and allowing Zapier to access your Google account. Then after have done that, you just select the document that you want this agent to learn and then click on Add Data Source. And you can see here you have a green dot here, which means it is all synced and the agent has learned that. Next, we have to configure our behaviors. Now, pro tip here, just add one task inside of each behavior because AI can still hallucinate. So the more specific the instruction that you can give it, the better the result that you are going to have. So to build our social media agent, we're actually going to put in 4 different behaviors. The first behavior will be to tell the agent to go ahead and do research about your niche, find the latest pieces of news. The second agent is going to go out and write the Facebook post. The third agent is going to write LinkedIn posts, and the fourth agent is going to write Instagram posts. The reason I don't bundle all these behaviors together is because people write differently on LinkedIn versus Instagram, and so I want the agent really know the nuances of that platform and write specific posts that's high converting for those platforms. I'm also going to include my prompts as well as the example post in the description below. So be sure to check that out. All right. So to create a new behavior, we're going to click on create behavior. And the trigger here is when we want this specific behavior to be performed. The first behavior we're going to do is the research agent, and we're going to use the schedule triggers. So basically, we want this agent to go out and do research about our niche on some sort of frequent basis. This could be daily, this could also be weekly. I'm going to do daily, and I can choose the time of day that I want this agent to go ahead and do this. Let's say it's 7 AM every morning, and I'm going to save this. Now in the instructions, and this is where you put in your prompt telling the agent what you want it to do. Okay, so I've put in my prompt for the research agent. And here's the instructions I have given to the AI. I said, your goal is to identify and compile a list of AI-related brands and companies that are being covered in the news online. So when you're using this prompt, which I'll include in the description below, feel free to just change this to whatever your niche may be. And I told it to conduct the research every day about news that was published yesterday. Okay, and then I told it to extract the following pieces of data: the date of the article, the title of the article, and the link to the article. And I also told it to summarize the article as well. Now I also want all of these news to be returned into a Google Sheet. This is what my Google Sheet looks like. So I told the AI to compile the information about the date, title, link, and the headline and the summary. And to link this Google Sheet now to this agent, I just had to press Insert Action up here, and then I will search for Google Sheet. And once I click on Google Sheet, I can see all of the actions that I can perform. And I actually want to create a new spreadsheet row. So basically, this research agent is going to go out to the web, search for news on the latest AI happenings, and then return that article into that spreadsheet that I just created. So once you have inserted this action, you have to link it to that spreadsheet that we had just created. Now to configure this action, First, you need to connect to your Google Sheets account, which I have already done so. Then for the subsequent fields, when we look at the dropdown, we can see that there's a few different options. We can let the agent select which drive or spreadsheet that we need to post this information in, or you can specify the specific value. I find that you're going to get more accurate results when you set a specific value in that field instead of letting the agent choose, because sometimes if you're like me, you have hundreds of files in your Google Drive, and it will be very hard for the agent to know which spreadsheet you wanted to post it in. For the spreadsheet, select the specific value for the field, and I've selected Zapier Agent Test, which is the spreadsheet I have created here. And for the worksheet, I selected Sheet 1. And then for all of the columns, so the date, title, link, this is where I let the agent generate those values. Okay, and I'm going to press save here. I have to save the entire thing, then I can go ahead and test this behavior. And this agent is now going to go do the work and take a look at the news that happened yesterday in the AI industry. So let's see what happens. Okay, we can see now it's searching Google, and to see what it has performed in each step, we can just expand on the task it's doing. Okay, so it's finished web browsing now. It's found the article, and now it's determining the next step. Oh, okay, so now it's creating that spreadsheet row. We can now go back to our spreadsheet, and we can see, boom, yes, it's got the date of the article, the title of the article, the link to the article, and it's summarized this piece of news for us as well. I mean, it's mind-blowing. I mean, social media manager is a job that companies routinely pay $30,000 to $60,000 a year for, for someone to perform this job for them. And now we can automate this whole thing with AI for like pennies on the dollar. You can have this whole automation set up and it would just be like $10 to $50 a month to run it. That's It's crazy how far we've gone with tech. All right, so once this has completed, we just have to turn on this action if it's performing the way that we want it to perform, and that's it. So now we're ready to do our next action. We're going to press the plus symbol here, and the next action will be to create our Facebook posting behavior. And for the trigger here, we are going to select Google Sheet. The trigger is what starts this specific behavior. It could be when a specific action happened in another app, or it's on schedule or on demand. So we're going to select Google Sheet for this trigger. Basically, whenever the research agent has finished doing its research, then we want it to take the latest research that the research agent has returned and write a unique high-converting Facebook post from there. So we're going to look for new spreadsheet row being added to our spreadsheet. And in the instructions, this is where we put in our prompt. All right, I have finished putting in my instructions for the Facebook agent. So I told it to write a high-engagement Facebook post that is designed to grab attention and spark conversation. I told it not to include any links and take a look at the article from the Google Sheet and use it to create this engaging Facebook post. Then I told it some instructions instructions like provide value, provide actionable steps, make it feel relevant and current, encourage discussions. And then I told it the tone and the style I wanted to use. And then here I can also insert some examples of posts so that the AI has more specific instructions of what I want it to do. The more specific the instructions that you can give to the AI, the better the result that you will get. Now I can actually insert in the data that we trained it on in the beginning now so that this agent has the examples of the posts that we wanted to use. So I just have to click on Insert Data here and sync it to the social media examples doc that I have. Now it's a great way to do it this way because you can just update that Google Doc and this agent's knowledge will automatically be updated. So it's a dynamic knowledge base versus a static knowledge base. After I have given it all the instructions, I also have to link this to the action. So I want it to automatically post onto my Facebook page once it's done. So to do that, I'm going to click on Insert Action, search for Facebook Pages, click on this, and look for Create New Page Post as the action. Once you insert in this action, be sure to configure this. You need to link this to your Facebook account. If you have multiple pages in your account, select the page you want it to insert. And for the message, photo, and link URL, I, I just let the agent generate those values. And I'm going to press Save and save this entire action. Now let's go ahead and test this behavior. Okay, so our social media agent is going to work now. It's taking a look at the spreadsheet that we just created. And it's also looking at the examples in the social media post. So it has loaded that content, and now it's thinking and summarizing the results. Great, now it has retrieved that content, and now it is crafting and writing the Facebook post for us. This is exciting. Okay, so it's been about a minute, and our AI agent has finished writing the post, and we can expand on this to see what it has written. I also went over to my my Facebook page here to read the post it has written. So basically, it summarized the article, picked out the most interesting pieces of AI news that happened yesterday, and wrote a short summary about it. Now, one thing I do want to show you, like, let's say you're not happy with this post, you can go ahead and change the instructions that you have given it. Or you can just tell the AI here in the reply box exactly what you don't like. Once we're happy with this behavior, we can just turn it on and then move on to the next one. So similarly, we're just going to use the exact same process to create our LinkedIn behavior and our Instagram behavior. Now I'm going to call this LinkedIn posting, and the trigger will again be the Google Sheet when a new spreadsheet row is added to our Zapier agent spreadsheet, and it'll be the first sheet. And in instructions, I'm going to tell it exactly how to write a high-converting LinkedIn post. I have put in my prompt for the LinkedIn behavior. I told it that you are a skilled LinkedIn content strategist. Your goal is to write a high-converting LinkedIn post that grabs attention, delivers value, and drives action. I told it the key elements that I want in the LinkedIn post, and then I go on to give it some examples of what I want this post to look like. And now I have to link it to my LinkedIn account here, so I'm going to insert action, search for LinkedIn, and go to LinkedIn. Then I want to select Create a Shared Update. Then I had to click on the action to configure it, link this to my LinkedIn account, and I'm going to let the agent choose the value of the comment field. So this is basically what's allowing the agent to post what it has just written. You can also choose whether you want this to be visible to everyone or just your connections, and I'm gonna the agent choose all of the content here. So once that is done, I just save this and press on test behavior again and wait a minute for this to be done. Okay, so it's been about a minute and the AI agent has finished crafting and posting my LinkedIn post here for me. If there's anything that you don't like about this post, just go back and forth in the reply section, tell it what you don't like, and then the AI will actually fix things up. And now let's say that I found the previous post here a little short and I want it to be longer, I can just in the reply comment here tell the AI to make it a bit longer. So I told it here to please make the post double the length and start a new line for each sentence. What I also want you to note here is that it can actually save your feedback. So it actually adds to your instructions as you're testing things. It actually adds to the instructions that you give it. It, you can see here that it added this line here by itself. So whatever feedback you're giving it, it's constantly taking that feedback and improving upon its results as well, which is super cool. Now it's creating a new LinkedIn post for me. All right, it's been about a minute or 2, and you can see the AI has written a new post that is an expanded version of what we had previously. So whatever you don't like, I mean, this is just super cool. Just type in your feedback And the agent can actually save the feedback by adding to your prompts. Now let's move on to our 4th and final behavior, which is creating our Instagram posting. The trigger is going to be the same one whenever there is a new spreadsheet row in the Zapier sheet. And now we have to put in our instructions. All right, so I have put in my instructions for the Instagram posting agent. So I have given it some instructions here. I told it the key elements I want in the captions. I gave it a role. I told the AI that you are a skilled Instagram content strategist. Your goal is to write engaging Instagram captions that grab attention, feels natural, and encourages likes, shares, and comments. Okay, I've given it some examples of captions. And lastly, here I told it to also generate an image. So let's add in that action. Now we're going to add in action again, and I'm going to use DALL-E to generate this image. So I'm going to search for OpenAI OpenAI. And then I'm going to search for generate an image with DALL-E. So currently, there are still some limitations with Zapier Agent. So you can't do a custom HTTP request, for example, that's not an action that can be done. So you can't link it to Flux or Stable Diffusion at this time. I hope it is a change that their team will make soon. And after you have connected this to your OpenAI account, you can let the agent choose the prompt. Prompt, the model that you want it to do. I'm going to choose DALL-E 3 for my model, and I am going to let the agent choose the number of images it wants to generate. And the rest looks good. Then lastly, once the caption as well as the image has been generated, I told it to link it to my Instagram account. So to do this, insert action, same thing, same, same. I'm going to search for Instagram for Business. And I'm going to select publish a photo. Then I'm going to click into it, link it to my account. If you have multiple accounts, then let the agent know which Instagram account you want to choose, and then let the agent choose the photo and the captions, and then press save. So once all that's done, let's test the behavior again and see what we get. It's looking at a new piece of news, and now it's generating the image. Okay, great. It's been about 2 minutes and we can see that the AI has generated an image for my Instagram account that is related to the news article it has found, and it has written a pretty good caption here as well. So if you don't have any kind of like social media presence, like do this, set this up, and then you have a completely automated system that will post on all the platforms forms for you. This is such a game changer. I hope you're starting to see the power of AI and automations. I hope you will give this a try. If you got anything out of this video, please like and subscribe. I also have a free AI automations course that you can sign up for as well. I'll also include the link to the Zapier agent and all of the prompts that we used in this video in the description below as well, so make sure you check that out. Thanks so much for joining me today, and I'll see you next time.