Exploring AI Automation Applications

Today, I am going to be bringing you 10, yes, 10 AI automation use cases that you can start implementing in order to save yourself time, make you more productive and efficient. Now, agentic systems is something you need to get ingrained in your head because that's all you are going to be hearing about over the next two years, 2025 and 2026. Not only is agent building very monetizable right now because of the massive content gap, not enough people know how to build automations and the demand is higher than it used to be. the actual supply. So you have the ability to make a lot of money if you know how to build these things. And if you can start getting yourself introduced to automations, but also for yourself, you know, I have so many automations that I use on a daily basis that make me smarter, that make me more productive, more efficient, save hours of my life, days of my year, saves my members of my community days of their year, members of my team. So I'm going to be sharing with you 10 of these automations in order to get you introduced. Now, agentic systems, like I said, are either fully autonomous agents that can act independently, or you can define agentic systems as prescriptive implementations with predefined outputs. Either way, we're working the same thing here. It's automation. And this is going to be a good starter video in order for you to get your foot in the door, implement some cool AI automations, see some ideas, and just watch how I am using them and other people around me. If everything I said just went over your head or you want a way to fast track your way to success, the AI Foundations community, my personal community along with my partner and brother Carter, is the spot you need to be. This is a community filled with people thinking about AI implementation all day long. We're talking about automations every day in here. We're talking about learning AI. We have a classroom with the basics of AI, like large language models, imagery, audio, and more. And then we have the actual implementation techniques in your categories like business, marketing, and sales. So in the business category, I have all of the automations that I'm going to be talking about in today's video already in the community with further implementation. I have our bulletproof systems all ready to go in a templated format. Now, this isn't a place to go grab a template and leave. This is a place to find your family, find your network of people, and stay. So if that interests you, I'll leave a link in the description and the top pinned comment below. We are welcoming you with open arms. When you join, make sure to introduce yourself because we want to get to know you. Automation use case number one is LLM split testing. Now, in Make.com, you have the ability to contact different large language models like JatGBT, Claude, Perplexity. You can even contact Gemini now, I believe. But with these large language models, not only can you contact their base model, But you have the ability to easily connect your API and you can select models within those large language models as well. So as you can see for this case, I'm just using 01 mini, but you can also select things like GPT-40, 01 preview, GPT-40 audio even. In this scenario, what I have is a form that's connected to a webhook. So whenever I type in a prompt in this form. It contacts a webhook in my Airtable input database. If you want to know how to build the input database, I've dedicated an entire YouTube video to building that for you guys. I'll leave that in the upper right-hand corner if you want to build that. But what we can do is I can just type out my prompt here, just as I would in ChatGPT or Cloud. And when I hit submit, it's going to send this prompt via a webhook to ChatGPT01mini, Cloud 3.5 Sonnet, Sonar Huge 128. k on perplexity and then it's going to update that record in Airtable. So when we submit a form it creates a record in my Airtable and it's going to bring it back to my categorized content under LLM split test. This prompt says describe this zoom call transcript in one word. Use one word in your response only. For example transformative. Here's the transcript. Now this prompt might need refining obviously in order to get the actual structured output but the idea of this is I can just you know copy and paste in my prompt to my LLM split test form I've created on Airtable, and then I can provide the transcript here at the end. So I'm going to go find a transcript that I have probably within my school database here. I'll just grab this transcript, I'll copy it. And then I will go back and this is an actual Zoom transcript from my AI Foundations community that I shared with you earlier. But I could paste in that transcript and then select that input type, a required field that we have to do for this to work and create an actual field. Then I can hit submit. And what you're going to notice is this creates a new form or excuse me, a new record within our Airtable database. So if I go to my categorized content. We now have a new record with the prompt in here, but we have no response. That response is going to run right here in this automation that I was just showing you. So as you can see, the information was just received. And by information, I mean the prompt from the form, you know, contacted this webhook with all of that data. And now it's sending it off to 01mini, sending it off to 3.5sonnet, sending it off to PerplexitySonar, huge model, and then updating that record in Airtable. So now we have... a response with this page that we just created. So we have our prompt here. We have the transcript, which is pretty long that I just pasted in that prompt. We have the 01 mini response. It says collaborative. We have chat. We have Claude 3.5 sonnet response, also collaborative. And perplexity says informative. So this is a good way to split test responses from different large language models. And it's very quick. I don't have to have open multiple browsers. I can just go to my form And I can even type off multiple prompts and just wait for them to populate within my database and check them later, you know? So great way to test. I could do this with 10 different models from the same company. I could do this from multiple different models from the same company and different companies. It doesn't really matter. The sky's the limit with Make.com and model testing. Automation use case number two is the PDF extraction system. So I wanted a simple way. to have AI scan the documents and the content on those PDF documents from a specific folder and provide me a structured summary with key takeaways. You know, some of these PDFs that I'm reading about artificial intelligence are, you know, 48 pages long, this one for instance, and it's very complex. You know, maybe before I even read this, I want to throw it through my PDF extraction system. That way I can have AI give me an executive summary, key takeaways, and a very structured output. So I have this folder in Google Drive that whenever I come across a new PDF that I enjoy or that I want to read, this is a relatively new system I'm implementing, I can throw it in this folder. On a schedule, I have make.com watching for new PDFs populating within that folder. When a new PDF populates in the folder, it's going to run through a PDF OCR, so reading all the content from the PDF, taking all that recognized text and sending it to AI. And what ChatGPT is going to do is give me structured PDF summaries in the form of JSON. So it's going to give me things like a title, executive summary, and key takeaways. These are just things that I wanted to provide because things that I thought were important. And you can get very specific, but as you can see, I have different prompts in here showing how I want the model to output its answer. Then I'm parsing the JSON and creating a record again in my input database in Airtable. I have a full video showing how to build that. It's a system AI Foundations uses. for many different automations, not all automations, of course, you can't throw everything into one database, but a lot of automations can fall back to this Airtable input database, you know, a lot of inputs coming in, but the structured outputs are what we're really after. So let me run through this automation just to show you real quick what it looks like. We have all those PDF files already within our Google Drive folder. So this is set up to watch that specific folder. If I hit... choose where to start, and then I just select all. So it watches all the PDFs and I hit run once. What this is going to do is run through that exact process that I just described. So it's found a file within the folder. It's ran through the PDF OCR. It's sent that information to chat GPT, and already it's created me an executive summary with key takeaways. And we have a 60 second pause for the tokens per minute input for chat GPT's API. So if we go back to Airtable now, as you can see, it already created a record. And it gives me, you know, the title of the document. If I open it up, it gives me an executive summary of what that specific document went over. It gives me key takeaways. You know, you can really structure the outputs. But the idea of this is we're taking PDFs running through OCR and getting structured outputs. You know, the structured output can be in whatever way you want, pretty much whatever channel you want. We had our client that we were working with actually use Microsoft SharePoint and they sent it to their Slack team. with their PDFs that they were uploading. Automation use case number three is this Zoom call input automation. Now there are so many different things you can do from Zoom calls. If you're a salesperson, if you're a community leader, if you're a team leader, whatever you are, whatever you Zoom for, I recommend implementing a system like this. What we have is such a seamless process happening in the background. It's absurd how actually easy this flows and how much time it saves us on a daily basis. In the AI Foundations community, we are doing three Zoom calls a week. What this automation is allowing us to do is it's watching recordings. So as soon as a recording gets done processing to the Zoom cloud, What happens is this automation begins without a schedule, without human intervention. This is webhook data. So it's actually getting responses. Okay, you know, this Zoom cloud recording is done. And what it does is it takes that call recording, downloads the transcript. This is what this first line is doing. It's checking in this filter if the recording file is a transcript or not. And it's taking that transcript and it's creating a record within our Airtable database. So that's like this record 27 here. within this Airtable database. It's taking it, it's creating a record, and uploading the transcript, the entire Zoom call transcript right here with labeled names and all that fun stuff. This next line, what's happening is it's going to download that mp4 file, the recorded Zoom call, upload it to Vimeo, which is our video software of choice, and then creating a record with that Vimeo link in Airtable. So again, If I go back to Airtable, what this is doing automatically right after we finish a Zoom call, no questions asked, you know, we go on about our day, but this stuff is happening for us in the background, is it's uploading this video to Vimeo from the Zoom cloud. So something you'd pay an assistant to do, an employee to do, it's very low cost compared to what you would have to pay somebody to do this. But if I click into this, as you can see, it's just our uploaded video, and it's just a video with our group. Beautiful. And then what's the last thing it's doing? Well, then it's taking the chat log from our zoom call, our community calls, and it's actually uploading and creating a record and air table around that as well. So it's uploading our entire chat here, which is very cool. Everyone's saying happy new year, very beautiful community that I'm blessed to be a part of. What you're going to maybe ask is why are we creating three different records? Well, that was just the easiest way I found to do it, you know, and automations are messy sometimes, but it's about getting the polished final result. So. With each one of these records comes a specific meeting ID. And we can actually capture records with the same meeting ID and then import it into one database item within a different Airtable database. So here in this two of two automation, we actually are taking those raw records like the transcript, the chat, the video. And what we can do is we can use AI in this case to give us a post name for our community so that when we're uploading call replays, the entire process is seamless from. the minute that we end the call to the entire post getting created, you know, all this is happening in the background. And it gives us a nice summary of what the call touched on. It gives us links that people were putting in the chat. It gives us the video URL that we can easily just copy and paste into our school community. I mean, this is a Zoom call automation that saves us literally days of our life because doing this process alone, like creating a summary, gathering the links. uploading the video, getting the transcript together. Takes probably about 45 minutes to an hour. So if we're doing three calls a week, we're saving three hours a week. The next automation is very exciting. It's the RSS feed input type automation. Now, this automation allows you to aggregate a bunch of sources, whether it's newsletters, news posts, blogs, YouTube videos, Instagram feeds, really whatever has a feed. You can put it. into a collection, into a bundle, and then run automations from it. So maybe I have an RSS feed, for example, that aggregates together all the news from these sources. My AI Foundation's YouTube channel, Artificial Intelligence on MIT News, AI News from artificialintelligencenews.com, OpenAI's research page, and TechCrunch's AI News and Artificial Intelligence category. RSS.app, which I'm on right now, allows you to put all of these feeds into one. So now all of those posts are funneling into this feed. We have an automation called the News Blast Automation that uses this RSS feed. And within our content generation system, we have it labeled RSS feeds. What it does is it actually gives a content summary. So it gives a title, the news article, and then the content from that page in the form of HTML that we could then use. We could use for our community. We could, you know, run through another AI automation that talks about these things. So as you can see, it just does a whole news blast over the past 48 hours. So, you know, I could run this if I wanted to. And when I run it, it's going to watch that feed for new articles. And it's actually watching a different feed that my brother set up for AI for his automated shorts automation. But as you can see, it creates a new record in our air table. So if I go back to the content generation system, a new record just populated right here. And this is just a huge news blast of up-to-date information. A very simple yet... effective automation. Learning RSS feeds for your automations is going to be very important. So important that I created a 30 minute tutorial in the AI foundations community in order to teach RSS feeds a little bit better. It wouldn't have done well on this channel. That's why I just uploaded that to the community. This next automation use case is honestly one of my favorites because of the simplicity for one and for two, how powerful it is. How many times do you want to grab the transcript from a YouTube video? You know, if you have those YouTube video transcripts, the amount of things you can do with them using AI is absolutely insane. You know, you can get key takeaways. You can actually get actionable steps from the text of what the creator spoke. So being able to have an automation setup that allows you to do that is crucial for your workflow. In my opinion, in today's day and age, especially if you're watching this video right now, you're obviously on YouTube. So if you had a way to scrape the transcripts, I'm sure that would be very powerful. So here's what I do. Maybe this can give you some ideas. In my Airtable, I have a form called YouTube URL to Transcript. And what this allows me to do is paste in a video URL. So this is what my form looks like. All I have to do is grab a video URL. Let's say I'm watching AI Foundations'video on if I had to relearn AI, I would do this. Great video. Highly recommend you watch it and take it serious. Not enough people are taking it seriously, but it's fine. I didn't expect or even want everyone to take it serious. But anyway, I'm going to grab this URL. Let's say I want the transcript. Instead of listening to myself talk, let's say I just want the actual spoken words. What I could do is I could grab that video URL, paste it in, select input type YouTube video and hit submit. That's all I have to do on the front end. In the back end, what's happening is this big automation in make.com, where, again, we're running an automation from Airtable so that when we hit submit on this form, it sends webhook data to make.com. And as you can see, it happens literally in the blink of an eye. But it runs through Apify Actors in order to scrape the transcript, gets the YouTube transcript, parses it, cleans it, and then... Creates a record in my air table database again. Everything's coming from the air table database. I love this thing So under my input type YouTube video, that's where I keep all of my YouTube video transcripts So one this automation is finished after it's done cleaning the transcript and fixing my improper speech with my lisp After it's done doing that then it updates the record and air table and we actually get a clean transcript What could I do with this transcript? Well as a content creator, these are words that I spoke myself in this YouTube video. So I'm taking my own video URLs and running them through this automation by inputting them in the form. And I could use this transcript in order to run it through another AI automation and create social media posts or create SLPs for my team or create structured guides for the AI foundations community. So as you can see, this automation is finished. If we go back to the input type YouTube video, it uploads in here with my title. and then the entire transcript. The next automation that I have for you is something similar to the YouTube video URL. But instead of pasting in a video URL, what you're going to do is type in an idea. And then AI is going to go out and do research for you on that idea and bring it back to your database. And everything stems from this idea input form. So if I open up this form, as soon as I input an idea and hit submit, it's going to send that idea to here. And then I'm using perplexity in order to do in-depth... research and white paper creation on whatever my idea I typed in here was. So you might be thinking, why don't you just do that in perplexity? Why do you have to do this in an automation? Well, I don't really feel like typing out this long prompt every single time. And maybe I want to customize this later down the line to have structured outputs, multi LLM reinforcement, maybe split testing. So it's very customizable and that's why I'm doing it this way. And I can just type off my ideas, submit and forget, you know? So maybe I have an idea here and I'll show this in action real quick of, I don't know, um, fasting for five days, you know, water only. I can just type out like a very basic idea and then it's going to go out and do like scientific research and white paper creation around this idea for me and bring it back into my categorized content within my air table database. So I'm going to just like basic text and submit. That's all I've got to do. I can go off about my day now. You know, AI is going to be working for me in the background. That's what we want it to do, right? Or I could just go in here and submit another idea. But as you can see, information has been received. It's doing in-depth research now that I'm using chat GPT to give my page and air table a name, because what you're going to notice is all of my air table input types have a name. So we have, as you can see, this one just populated in from the automation that just got complete. But that's what ChatGPT is doing is giving this an idea name. So it named it Water Fasting Benefits. And then we have the raw text of the deep research it went into for me on this topic. So maybe I don't want to do it after reading this, or maybe I want to go into it even more. And again, the structured outputs are entirely up to you. I just wanted to show you an idea of what you can do. going from basic text to deep research in a matter of seconds, especially if you have long prompts that you like to use. I like keeping those in the developer make side of things. The next automation I want to show you is very powerful because it's activated by your voice. So I have a Slack channel and this Slack channel is where I keep all my ideas, let's just say. And I haven't used this channel in specific in a while, but what I can do is I can just record me talking. And I have an automation setup that's going to watch for audio in this specific Slack channel, download that audio, transcribe audio to text, give that text idea a name, and create a Airtable page for me. Now, I keep creating Airtable pages. You could send yourself emails. You could send yourself text messages, notifications to your phone, whatever you want. You know, automations just need to start somewhere and end somewhere. A lot of the time, I prefer ending them and starting them in Airtable. But in this case, we're doing Slack. So I'll do a live example for you here real quick, just to show you what I'm talking about. I could do this and I could hit record audio clip. Again, I could do this on my phone. I think that many people misunderstand how to actually use AI. I think that there are a lot of distractions out there. And I think in order for people to be successful, they need to master four sectors of AI. And then they need to learn automations, join a network, and of course, join the AI Foundation's community. But now that I have this... you know, idea in my Slack channel, this automation could be running on a schedule every one minute, two minutes, three minutes, hour, 24 hours, 48, doesn't matter. I'm just going to hit run once for the sake of this video. And it's going to download that audio file, create an air table page of my dictated speech. So now we have it right in my air table and it's word for word what I just said live on camera. The next automation I've created doesn't really use make.com, but it involves very similar principles. It actually uses my iPhone. What this allows me to do is tap a button on my iPhone and then actually get the dictated speech into my Airtable database. So I'm going to show you a live example of what that looks like. Again, I have all these automations in the community, the AI Foundations community, where I walk you through them, give you our system, and then we have calls for support as well. But Let me show you how this is working. So I have my iPhone here. I can just tap this widget on my phone. I have an idea, or I can just tell Siri, I have an idea. So for now I'll just tap it. I have an idea and then I'll speak into it. Hi there. Just wanted to show these awesome viewers who've stuck out this far, um, this automation that I've created with Siri. And when I hit stop, what that's going to do is use chat GPT in the background of the, uh, iPhone shortcuts, and it's going to give it a name. What this will also do when I hit done on my iPhone is upload it to my inputs right here. So as you can see, the last one was January 2nd. If I hit done on my iPhone, which I will do right now, that's going to automatically upload my dictated speech to my input database. We're going to wait a few seconds to watch it come in. And this would be an example of, you know, A quick automation that could save you a lot of time. If you ever have ideas, it's great that you can use your phone as a tool for you. It's a great tool to have. As you can see, it just uploaded in here and we have my exact word for word dictated speech. I could even have another automation that scrapes the Siri input types and creates business plans from them, extracts important information. I could set it up to have full conversations with me and my brother, whatever you want. But... That's another automation I use on a daily basis in order to save me time. Now this next automation, I will admit, is a little bit personal. But what I have is a ring. And this ring tracks my sleep. Many of you have probably heard of it. It is called the Oura Ring. And it has sensors in it, can track body temperature. I was just recently sick, just coming off a sickness actually. And this thing basically warned me that I was sick before I knew that I was getting sick. Okay, you have a thing called a readiness score. And my readiness score was dropping because of some of my metrics, like my heart rate variability, my temperature while I sleep, my breathing, respiratory rate. So what I have is an automation in here in order to help me track some of my health, because this ring has an API. But what I have is this automation importing my sleep score, my readiness data, everything I just talked about. Into my personal notion dashboard. Now, this was created by my brother Carter, otherwise known as Productive Dude, otherwise known as co-founder of AI Foundations. And he's an absolute wizard with the personal dashboards, notion, AI, literally, you know, pretty much everything. But what we have here is my personal database where I go over today's tasks, my notes, this is where I store all my goals for the year. Literally everything. This thing keeps me on track. But what I also have... It's this automation importing my sleep stats once a day at 12 p.m. I'm going to run this one a little bit early so you guys can see what it looks like in action. But this is going to aggregate all my sleep data to my Notion account so I can see everything within my daily habits. So if I hit run once, what this is going to do is pull all of my last night's sleep stats, my readiness score, and everything you can imagine. So now when I go back to my digital brain, as you can see, it has the page automatically created for me, and it fills out... all of my sleep stats. So yes, a very kind of light night of sleep. You know, I only slept five hours and 54 minutes last night, but this just takes the thinking out of this. I don't even have to go to the Oura Ring app anymore. I can just limit my tools that I use because it aggregates everything into this Notion page for me. It gives me my sleep score, gives me how long I slept for certain stages of the night, you know, my deep sleep, my light sleep, my REM sleep, my total sleep, sleep efficiency, gives me a nice little score right here. My body temperature was negative 0.23. So you guys know pretty much everything about me and what happened last night. But this is a way and an example of how you can actually use technology in order to gain deeper insights about yourself. Once I gather enough data, I could export my Notion database as a CSV and my journal entries as well. And I could upload it and I could start having AI draw connections between my sentiment and my journal entries, my sleep score. how much work I got done in tasks, because this is my task database as well, all the connections start coming together. That's how you could use it for personal well-being, personal health, personal insight. It doesn't really just have to be for business, business, business, right? I like using this stuff for personal as well. The last automation I wanted to show you guys involves HTTP request from your GPTs and chat GPT. This is something that a lot of people come to me with questions for, and also that I use. Whenever I want to retrieve information and send information all from a GPT hub. Let's say that in chat GPT, I like creating blogs, but I want to create blogs from my Slack recording inputs in my input database. So whenever I speak a message into Slack, I already showed you that automation. It gets imported into this database here in the form of text, my actual spoken text. Let's say from my spoken text, I like creating blog posts. Well, in order to actually have a good workflow, what I can do is I can set up an HTTP request in the schema of my GPT. And what this allows me to do is pull content from other websites, such as my Airtable, and automatically have that pull into here without me even copying and pasting anything. And I can do that by setting up an HTTP request and by adding actions and schema to this GPT. If I go into the backend here and hit edit GPT, what you're going to notice is I have a couple of instructions here. But I also have an action using a webhook with make.com. So anytime I contact this webhook, what it's going to do is search my Airtable for specific records with the input type of Slack recording. If I click in here, that's why I like labeling all my Airtable records with input types is because I can search records with very specific input types. So let's say I just want to pull in Slack recordings. I can do that by just searching for this input type. and then aggregating all those records and then sending them off to my GPT with a webhook response. Now on the back end, that looks like this, right? I have my OpenAI schema here with an available action of a get request. Very simple. I can fetch idea details. And this is more complex, but this is just an example of how we're using automations. Maybe I have an idea that I want to create a blog post around. Well, I could start that in Slack. Maybe I'm on the go. Maybe I'm walking, maybe I'm cooking, whatever. Maybe an idea comes to mind. I could open up Slack on my phone or in this case on my computer and I could speak. I think more people should use ChatGPT in order to help them with creative writing and business plans. I think the combination of creative writing that ChatGPT offers along with the business plan expertise could be a recipe for people actually getting work done and actually doing more valuable tasks. Remember this automation that I showed you? So when we hit run once, it's actually going to take that audio file that I just recorded and upload it to my Airtable database. As you see, it just populated in here. So now at the bottom, we have this exact word for word thing, what I spoke with you here live. Now what I can do with that GPT actions that I was showing you was say fetch. or start or go. Doesn't really matter. You can set it up however you want. And then it's going to talk to make.com and contact this automation that I set up. And then it's going to put all of my air table fields, all of my inputs in a table for me in chat GPT without having to lift a finger off chat GPT. Then it says, let me know which idea you'd like to turn into a blog post. So this is dynamically updating as my. Airtable database updates as well. So those are 10 AI automation use cases in order to get your head spinning on how you can start implementing AI and building agentic systems in order to make your life more productive and efficient. Again, if you want to learn more about each one of these systems, you want to download templates, you want to install them, you want to be with a family and a band of people that actually understand the stuff that talk about it all day long, then the AI Foundations community is where you need to be. And again, And 80% of the automations that I showed you, you know, eight of 10 of them are within our input database system using AI for business. I have more modules going very in depth on each one of these automations that I showed you today, how to build them. And once you understand the how, you can start really building automations for yourself. And that's kind of the goal of this community is to get you to learn how to build automations, agentic systems for yourself. So I highly recommend you join a network of people. like the AI Foundations community, and get started building with us. Now, I hope this video was helpful regardless today. I hope that you learned something new, got a new idea. If you have any questions, please leave them in the comments or join our free AI community, AI Pioneers. I'll leave that in the description as well. And with that being said, I hope you enjoyed. Please subscribe and like this video if you want more content like this, and I will see you in the next one.

So you have the ability to make a lot of money if you know how to build these things. And if you can start getting yourself introduced to automations, but also for yourself, you know, I have so many automations that I use on a daily basis that make me smarter, that make me more productive, more efficient, save hours of my life, days of my year, saves my members of my community days of their year, members of my team. So I'm going to be sharing with you 10 of these automations in order to get you introduced. Now, agentic systems, like I said, are either fully autonomous agents that can act independently, or you can define agentic systems as prescriptive implementations with predefined outputs. Either way, we're working the same thing here.

It's automation. And this is going to be a good starter video in order for you to get your foot in the door, implement some cool AI automations, see some ideas, and just watch how I am using them and other people around me. If everything I said just went over your head or you want a way to fast track your way to success, the AI Foundations community, my personal community along with my partner and brother Carter, is the spot you need to be.

This is a community filled with people thinking about AI implementation all day long. We're talking about automations every day in here. We're talking about learning AI. We have a classroom with the basics of AI, like large language models, imagery, audio, and more.

And then we have the actual implementation techniques in your categories like business, marketing, and sales. So in the business category, I have all of the automations that I'm going to be talking about in today's video already in the community with further implementation. I have our bulletproof systems all ready to go in a templated format.

Now, this isn't a place to go grab a template and leave. This is a place to find your family, find your network of people, and stay. So if that interests you, I'll leave a link in the description and the top pinned comment below.

We are welcoming you with open arms. When you join, make sure to introduce yourself because we want to get to know you. Automation use case number one is LLM split testing.

Now, in Make.com, you have the ability to contact different large language models like JatGBT, Claude, Perplexity. You can even contact Gemini now, I believe. But with these large language models, not only can you contact their base model, But you have the ability to easily connect your API and you can select models within those large language models as well. So as you can see for this case, I'm just using 01 mini, but you can also select things like GPT-40, 01 preview, GPT-40 audio even.

In this scenario, what I have is a form that's connected to a webhook. So whenever I type in a prompt in this form. It contacts a webhook in my Airtable input database.

If you want to know how to build the input database, I've dedicated an entire YouTube video to building that for you guys. I'll leave that in the upper right-hand corner if you want to build that. But what we can do is I can just type out my prompt here, just as I would in ChatGPT or Cloud.

And when I hit submit, it's going to send this prompt via a webhook to ChatGPT01mini, Cloud 3.5 Sonnet, Sonar Huge 128. k on perplexity and then it's going to update that record in Airtable. So when we submit a form it creates a record in my Airtable and it's going to bring it back to my categorized content under LLM split test. This prompt says describe this zoom call transcript in one word.

Use one word in your response only. For example transformative. Here's the transcript. Now this prompt might need refining obviously in order to get the actual structured output but the idea of this is I can just you know copy and paste in my prompt to my LLM split test form I've created on Airtable, and then I can provide the transcript here at the end. So I'm going to go find a transcript that I have probably within my school database here.

I'll just grab this transcript, I'll copy it. And then I will go back and this is an actual Zoom transcript from my AI Foundations community that I shared with you earlier. But I could paste in that transcript and then select that input type, a required field that we have to do for this to work and create an actual field. Then I can hit submit.

And what you're going to notice is this creates a new form or excuse me, a new record within our Airtable database. So if I go to my categorized content. We now have a new record with the prompt in here, but we have no response.

That response is going to run right here in this automation that I was just showing you. So as you can see, the information was just received. And by information, I mean the prompt from the form, you know, contacted this webhook with all of that data.

And now it's sending it off to 01mini, sending it off to 3.5sonnet, sending it off to PerplexitySonar, huge model, and then updating that record in Airtable. So now we have... a response with this page that we just created.

So we have our prompt here. We have the transcript, which is pretty long that I just pasted in that prompt. We have the 01 mini response. It says collaborative.

We have chat. We have Claude 3.5 sonnet response, also collaborative. And perplexity says informative.

So this is a good way to split test responses from different large language models. And it's very quick. I don't have to have open multiple browsers.

I can just go to my form And I can even type off multiple prompts and just wait for them to populate within my database and check them later, you know? So great way to test. I could do this with 10 different models from the same company. I could do this from multiple different models from the same company and different companies. It doesn't really matter.

The sky's the limit with Make.com and model testing. Automation use case number two is the PDF extraction system. So I wanted a simple way.

to have AI scan the documents and the content on those PDF documents from a specific folder and provide me a structured summary with key takeaways. You know, some of these PDFs that I'm reading about artificial intelligence are, you know, 48 pages long, this one for instance, and it's very complex. You know, maybe before I even read this, I want to throw it through my PDF extraction system. That way I can have AI give me an executive summary, key takeaways, and a very structured output. So I have this folder in Google Drive that whenever I come across a new PDF that I enjoy or that I want to read, this is a relatively new system I'm implementing, I can throw it in this folder.

On a schedule, I have make.com watching for new PDFs populating within that folder. When a new PDF populates in the folder, it's going to run through a PDF OCR, so reading all the content from the PDF, taking all that recognized text and sending it to AI. And what ChatGPT is going to do is give me structured PDF summaries in the form of JSON.

So it's going to give me things like a title, executive summary, and key takeaways. These are just things that I wanted to provide because things that I thought were important. And you can get very specific, but as you can see, I have different prompts in here showing how I want the model to output its answer. Then I'm parsing the JSON and creating a record again in my input database in Airtable.

I have a full video showing how to build that. It's a system AI Foundations uses. for many different automations, not all automations, of course, you can't throw everything into one database, but a lot of automations can fall back to this Airtable input database, you know, a lot of inputs coming in, but the structured outputs are what we're really after. So let me run through this automation just to show you real quick what it looks like.

We have all those PDF files already within our Google Drive folder. So this is set up to watch that specific folder. If I hit... choose where to start, and then I just select all. So it watches all the PDFs and I hit run once.

What this is going to do is run through that exact process that I just described. So it's found a file within the folder. It's ran through the PDF OCR. It's sent that information to chat GPT, and already it's created me an executive summary with key takeaways. And we have a 60 second pause for the tokens per minute input for chat GPT's API.

So if we go back to Airtable now, as you can see, it already created a record. And it gives me, you know, the title of the document. If I open it up, it gives me an executive summary of what that specific document went over.

It gives me key takeaways. You know, you can really structure the outputs. But the idea of this is we're taking PDFs running through OCR and getting structured outputs.

You know, the structured output can be in whatever way you want, pretty much whatever channel you want. We had our client that we were working with actually use Microsoft SharePoint and they sent it to their Slack team. with their PDFs that they were uploading. Automation use case number three is this Zoom call input automation.

Now there are so many different things you can do from Zoom calls. If you're a salesperson, if you're a community leader, if you're a team leader, whatever you are, whatever you Zoom for, I recommend implementing a system like this. What we have is such a seamless process happening in the background.

It's absurd how actually easy this flows and how much time it saves us on a daily basis. In the AI Foundations community, we are doing three Zoom calls a week. What this automation is allowing us to do is it's watching recordings.

So as soon as a recording gets done processing to the Zoom cloud, What happens is this automation begins without a schedule, without human intervention. This is webhook data. So it's actually getting responses.

Okay, you know, this Zoom cloud recording is done. And what it does is it takes that call recording, downloads the transcript. This is what this first line is doing. It's checking in this filter if the recording file is a transcript or not.

And it's taking that transcript and it's creating a record within our Airtable database. So that's like this record 27 here. within this Airtable database.

It's taking it, it's creating a record, and uploading the transcript, the entire Zoom call transcript right here with labeled names and all that fun stuff. This next line, what's happening is it's going to download that mp4 file, the recorded Zoom call, upload it to Vimeo, which is our video software of choice, and then creating a record with that Vimeo link in Airtable. So again, If I go back to Airtable, what this is doing automatically right after we finish a Zoom call, no questions asked, you know, we go on about our day, but this stuff is happening for us in the background, is it's uploading this video to Vimeo from the Zoom cloud.

So something you'd pay an assistant to do, an employee to do, it's very low cost compared to what you would have to pay somebody to do this. But if I click into this, as you can see, it's just our uploaded video, and it's just a video with our group. Beautiful. And then what's the last thing it's doing? Well, then it's taking the chat log from our zoom call, our community calls, and it's actually uploading and creating a record and air table around that as well.

So it's uploading our entire chat here, which is very cool. Everyone's saying happy new year, very beautiful community that I'm blessed to be a part of. What you're going to maybe ask is why are we creating three different records?

Well, that was just the easiest way I found to do it, you know, and automations are messy sometimes, but it's about getting the polished final result. So. With each one of these records comes a specific meeting ID. And we can actually capture records with the same meeting ID and then import it into one database item within a different Airtable database.

So here in this two of two automation, we actually are taking those raw records like the transcript, the chat, the video. And what we can do is we can use AI in this case to give us a post name for our community so that when we're uploading call replays, the entire process is seamless from. the minute that we end the call to the entire post getting created, you know, all this is happening in the background. And it gives us a nice summary of what the call touched on.

It gives us links that people were putting in the chat. It gives us the video URL that we can easily just copy and paste into our school community. I mean, this is a Zoom call automation that saves us literally days of our life because doing this process alone, like creating a summary, gathering the links.

uploading the video, getting the transcript together. Takes probably about 45 minutes to an hour. So if we're doing three calls a week, we're saving three hours a week. The next automation is very exciting.

It's the RSS feed input type automation. Now, this automation allows you to aggregate a bunch of sources, whether it's newsletters, news posts, blogs, YouTube videos, Instagram feeds, really whatever has a feed. You can put it.

into a collection, into a bundle, and then run automations from it. So maybe I have an RSS feed, for example, that aggregates together all the news from these sources. My AI Foundation's YouTube channel, Artificial Intelligence on MIT News, AI News from artificialintelligencenews.com, OpenAI's research page, and TechCrunch's AI News and Artificial Intelligence category.

RSS.app, which I'm on right now, allows you to put all of these feeds into one. So now all of those posts are funneling into this feed. We have an automation called the News Blast Automation that uses this RSS feed. And within our content generation system, we have it labeled RSS feeds.

What it does is it actually gives a content summary. So it gives a title, the news article, and then the content from that page in the form of HTML that we could then use. We could use for our community.

We could, you know, run through another AI automation that talks about these things. So as you can see, it just does a whole news blast over the past 48 hours. So, you know, I could run this if I wanted to.

And when I run it, it's going to watch that feed for new articles. And it's actually watching a different feed that my brother set up for AI for his automated shorts automation. But as you can see, it creates a new record in our air table. So if I go back to the content generation system, a new record just populated right here. And this is just a huge news blast of up-to-date information.

A very simple yet... effective automation. Learning RSS feeds for your automations is going to be very important. So important that I created a 30 minute tutorial in the AI foundations community in order to teach RSS feeds a little bit better.

It wouldn't have done well on this channel. That's why I just uploaded that to the community. This next automation use case is honestly one of my favorites because of the simplicity for one and for two, how powerful it is.

How many times do you want to grab the transcript from a YouTube video? You know, if you have those YouTube video transcripts, the amount of things you can do with them using AI is absolutely insane. You know, you can get key takeaways. You can actually get actionable steps from the text of what the creator spoke.

So being able to have an automation setup that allows you to do that is crucial for your workflow. In my opinion, in today's day and age, especially if you're watching this video right now, you're obviously on YouTube. So if you had a way to scrape the transcripts, I'm sure that would be very powerful.

So here's what I do. Maybe this can give you some ideas. In my Airtable, I have a form called YouTube URL to Transcript.

And what this allows me to do is paste in a video URL. So this is what my form looks like. All I have to do is grab a video URL.

Let's say I'm watching AI Foundations'video on if I had to relearn AI, I would do this. Great video. Highly recommend you watch it and take it serious. Not enough people are taking it seriously, but it's fine.

I didn't expect or even want everyone to take it serious. But anyway, I'm going to grab this URL. Let's say I want the transcript. Instead of listening to myself talk, let's say I just want the actual spoken words.

What I could do is I could grab that video URL, paste it in, select input type YouTube video and hit submit. That's all I have to do on the front end. In the back end, what's happening is this big automation in make.com, where, again, we're running an automation from Airtable so that when we hit submit on this form, it sends webhook data to make.com. And as you can see, it happens literally in the blink of an eye.

But it runs through Apify Actors in order to scrape the transcript, gets the YouTube transcript, parses it, cleans it, and then... Creates a record in my air table database again. Everything's coming from the air table database. I love this thing So under my input type YouTube video, that's where I keep all of my YouTube video transcripts So one this automation is finished after it's done cleaning the transcript and fixing my improper speech with my lisp After it's done doing that then it updates the record and air table and we actually get a clean transcript What could I do with this transcript? Well as a content creator, these are words that I spoke myself in this YouTube video.

So I'm taking my own video URLs and running them through this automation by inputting them in the form. And I could use this transcript in order to run it through another AI automation and create social media posts or create SLPs for my team or create structured guides for the AI foundations community. So as you can see, this automation is finished.

If we go back to the input type YouTube video, it uploads in here with my title. and then the entire transcript. The next automation that I have for you is something similar to the YouTube video URL.

But instead of pasting in a video URL, what you're going to do is type in an idea. And then AI is going to go out and do research for you on that idea and bring it back to your database. And everything stems from this idea input form. So if I open up this form, as soon as I input an idea and hit submit, it's going to send that idea to here.

And then I'm using perplexity in order to do in-depth... research and white paper creation on whatever my idea I typed in here was. So you might be thinking, why don't you just do that in perplexity?

Why do you have to do this in an automation? Well, I don't really feel like typing out this long prompt every single time. And maybe I want to customize this later down the line to have structured outputs, multi LLM reinforcement, maybe split testing.

So it's very customizable and that's why I'm doing it this way. And I can just type off my ideas, submit and forget, you know? So maybe I have an idea here and I'll show this in action real quick of, I don't know, um, fasting for five days, you know, water only. I can just type out like a very basic idea and then it's going to go out and do like scientific research and white paper creation around this idea for me and bring it back into my categorized content within my air table database.

So I'm going to just like basic text and submit. That's all I've got to do. I can go off about my day now.

You know, AI is going to be working for me in the background. That's what we want it to do, right? Or I could just go in here and submit another idea. But as you can see, information has been received.

It's doing in-depth research now that I'm using chat GPT to give my page and air table a name, because what you're going to notice is all of my air table input types have a name. So we have, as you can see, this one just populated in from the automation that just got complete. But that's what ChatGPT is doing is giving this an idea name. So it named it Water Fasting Benefits.

And then we have the raw text of the deep research it went into for me on this topic. So maybe I don't want to do it after reading this, or maybe I want to go into it even more. And again, the structured outputs are entirely up to you.

I just wanted to show you an idea of what you can do. going from basic text to deep research in a matter of seconds, especially if you have long prompts that you like to use. I like keeping those in the developer make side of things.

The next automation I want to show you is very powerful because it's activated by your voice. So I have a Slack channel and this Slack channel is where I keep all my ideas, let's just say. And I haven't used this channel in specific in a while, but what I can do is I can just record me talking.

And I have an automation setup that's going to watch for audio in this specific Slack channel, download that audio, transcribe audio to text, give that text idea a name, and create a Airtable page for me. Now, I keep creating Airtable pages. You could send yourself emails.

You could send yourself text messages, notifications to your phone, whatever you want. You know, automations just need to start somewhere and end somewhere. A lot of the time, I prefer ending them and starting them in Airtable.

But in this case, we're doing Slack. So I'll do a live example for you here real quick, just to show you what I'm talking about. I could do this and I could hit record audio clip.

Again, I could do this on my phone. I think that many people misunderstand how to actually use AI. I think that there are a lot of distractions out there.

And I think in order for people to be successful, they need to master four sectors of AI. And then they need to learn automations, join a network, and of course, join the AI Foundation's community. But now that I have this... you know, idea in my Slack channel, this automation could be running on a schedule every one minute, two minutes, three minutes, hour, 24 hours, 48, doesn't matter. I'm just going to hit run once for the sake of this video.

And it's going to download that audio file, create an air table page of my dictated speech. So now we have it right in my air table and it's word for word what I just said live on camera. The next automation I've created doesn't really use make.com, but it involves very similar principles.

It actually uses my iPhone. What this allows me to do is tap a button on my iPhone and then actually get the dictated speech into my Airtable database. So I'm going to show you a live example of what that looks like.

Again, I have all these automations in the community, the AI Foundations community, where I walk you through them, give you our system, and then we have calls for support as well. But Let me show you how this is working. So I have my iPhone here.

I can just tap this widget on my phone. I have an idea, or I can just tell Siri, I have an idea. So for now I'll just tap it. I have an idea and then I'll speak into it. Hi there.

Just wanted to show these awesome viewers who've stuck out this far, um, this automation that I've created with Siri. And when I hit stop, what that's going to do is use chat GPT in the background of the, uh, iPhone shortcuts, and it's going to give it a name. What this will also do when I hit done on my iPhone is upload it to my inputs right here. So as you can see, the last one was January 2nd. If I hit done on my iPhone, which I will do right now, that's going to automatically upload my dictated speech to my input database.

We're going to wait a few seconds to watch it come in. And this would be an example of, you know, A quick automation that could save you a lot of time. If you ever have ideas, it's great that you can use your phone as a tool for you.

It's a great tool to have. As you can see, it just uploaded in here and we have my exact word for word dictated speech. I could even have another automation that scrapes the Siri input types and creates business plans from them, extracts important information. I could set it up to have full conversations with me and my brother, whatever you want. But...

That's another automation I use on a daily basis in order to save me time. Now this next automation, I will admit, is a little bit personal. But what I have is a ring. And this ring tracks my sleep.

Many of you have probably heard of it. It is called the Oura Ring. And it has sensors in it, can track body temperature. I was just recently sick, just coming off a sickness actually.

And this thing basically warned me that I was sick before I knew that I was getting sick. Okay, you have a thing called a readiness score. And my readiness score was dropping because of some of my metrics, like my heart rate variability, my temperature while I sleep, my breathing, respiratory rate. So what I have is an automation in here in order to help me track some of my health, because this ring has an API. But what I have is this automation importing my sleep score, my readiness data, everything I just talked about.

Into my personal notion dashboard. Now, this was created by my brother Carter, otherwise known as Productive Dude, otherwise known as co-founder of AI Foundations. And he's an absolute wizard with the personal dashboards, notion, AI, literally, you know, pretty much everything. But what we have here is my personal database where I go over today's tasks, my notes, this is where I store all my goals for the year. Literally everything.

This thing keeps me on track. But what I also have... It's this automation importing my sleep stats once a day at 12 p.m.

I'm going to run this one a little bit early so you guys can see what it looks like in action. But this is going to aggregate all my sleep data to my Notion account so I can see everything within my daily habits. So if I hit run once, what this is going to do is pull all of my last night's sleep stats, my readiness score, and everything you can imagine. So now when I go back to my digital brain, as you can see, it has the page automatically created for me, and it fills out... all of my sleep stats.

So yes, a very kind of light night of sleep. You know, I only slept five hours and 54 minutes last night, but this just takes the thinking out of this. I don't even have to go to the Oura Ring app anymore. I can just limit my tools that I use because it aggregates everything into this Notion page for me.

It gives me my sleep score, gives me how long I slept for certain stages of the night, you know, my deep sleep, my light sleep, my REM sleep, my total sleep, sleep efficiency, gives me a nice little score right here. My body temperature was negative 0.23. So you guys know pretty much everything about me and what happened last night.

But this is a way and an example of how you can actually use technology in order to gain deeper insights about yourself. Once I gather enough data, I could export my Notion database as a CSV and my journal entries as well. And I could upload it and I could start having AI draw connections between my sentiment and my journal entries, my sleep score.

how much work I got done in tasks, because this is my task database as well, all the connections start coming together. That's how you could use it for personal well-being, personal health, personal insight. It doesn't really just have to be for business, business, business, right? I like using this stuff for personal as well.

The last automation I wanted to show you guys involves HTTP request from your GPTs and chat GPT. This is something that a lot of people come to me with questions for, and also that I use. Whenever I want to retrieve information and send information all from a GPT hub. Let's say that in chat GPT, I like creating blogs, but I want to create blogs from my Slack recording inputs in my input database.

So whenever I speak a message into Slack, I already showed you that automation. It gets imported into this database here in the form of text, my actual spoken text. Let's say from my spoken text, I like creating blog posts. Well, in order to actually have a good workflow, what I can do is I can set up an HTTP request in the schema of my GPT. And what this allows me to do is pull content from other websites, such as my Airtable, and automatically have that pull into here without me even copying and pasting anything.

And I can do that by setting up an HTTP request and by adding actions and schema to this GPT. If I go into the backend here and hit edit GPT, what you're going to notice is I have a couple of instructions here. But I also have an action using a webhook with make.com.

So anytime I contact this webhook, what it's going to do is search my Airtable for specific records with the input type of Slack recording. If I click in here, that's why I like labeling all my Airtable records with input types is because I can search records with very specific input types. So let's say I just want to pull in Slack recordings.

I can do that by just searching for this input type. and then aggregating all those records and then sending them off to my GPT with a webhook response. Now on the back end, that looks like this, right? I have my OpenAI schema here with an available action of a get request. Very simple.

I can fetch idea details. And this is more complex, but this is just an example of how we're using automations. Maybe I have an idea that I want to create a blog post around.

Well, I could start that in Slack. Maybe I'm on the go. Maybe I'm walking, maybe I'm cooking, whatever. Maybe an idea comes to mind. I could open up Slack on my phone or in this case on my computer and I could speak.

I think more people should use ChatGPT in order to help them with creative writing and business plans. I think the combination of creative writing that ChatGPT offers along with the business plan expertise could be a recipe for people actually getting work done and actually doing more valuable tasks. Remember this automation that I showed you? So when we hit run once, it's actually going to take that audio file that I just recorded and upload it to my Airtable database.

As you see, it just populated in here. So now at the bottom, we have this exact word for word thing, what I spoke with you here live. Now what I can do with that GPT actions that I was showing you was say fetch. or start or go.

Doesn't really matter. You can set it up however you want. And then it's going to talk to make.com and contact this automation that I set up. And then it's going to put all of my air table fields, all of my inputs in a table for me in chat GPT without having to lift a finger off chat GPT.

Then it says, let me know which idea you'd like to turn into a blog post. So this is dynamically updating as my. Airtable database updates as well. So those are 10 AI automation use cases in order to get your head spinning on how you can start implementing AI and building agentic systems in order to make your life more productive and efficient. Again, if you want to learn more about each one of these systems, you want to download templates, you want to install them, you want to be with a family and a band of people that actually understand the stuff that talk about it all day long, then the AI Foundations community is where you need to be.

And again, And 80% of the automations that I showed you, you know, eight of 10 of them are within our input database system using AI for business. I have more modules going very in depth on each one of these automations that I showed you today, how to build them. And once you understand the how, you can start really building automations for yourself.

And that's kind of the goal of this community is to get you to learn how to build automations, agentic systems for yourself. So I highly recommend you join a network of people. like the AI Foundations community, and get started building with us. Now, I hope this video was helpful regardless today.

I hope that you learned something new, got a new idea. If you have any questions, please leave them in the comments or join our free AI community, AI Pioneers. I'll leave that in the description as well. And with that being said, I hope you enjoyed.

Please subscribe and like this video if you want more content like this, and I will see you in the next one.

Transcript for:Exploring AI Automation Applications

Transcript for:
Exploring AI Automation Applications