I'm Giving You My Favorite Business Automation For Free

Today I have the pleasure of showing you one of my personal favorite automations. And I mean that. We try to build a lot of different automations to support the content and community business that we run, but not all of them stick. But one of the big ones that we not just keep around, but keep iterating on over time is this meeting summary automation right here. And today I will be guiding you through this automation, giving you the blueprint and telling you what possibilities you kind of have here. But first, let me explain what this is. It basically takes in brand new meeting notes from our team meetings and then runs a GPT API call on top of it and then basically takes the content of the meeting and extracts valuable insights for me as the CEO and founder of our little organization here. Ultimately, this is what that looks like. It doesn't just give me the action items, which is pretty standard, but it also extracts custom ideas that are aligned with my own goals. And it also surfaces uncomfortable truths from within the transcripts. These are custom prompting techniques that are not present in any of the meeting summarization tools I'm aware of, and I really wanted this. As you know, by now, I have a large collection of various specialty prompts. But it's unrealistic to just pull every meeting transcript and do this manually. That's why we created this automation that does all of this. And it even has an element in there where it shows how well that meeting aligned with our current road map. But as you can see, this is quite custom and I will do my best to explain it to you in this video and you will get the blueprint. But I just have to say that this is a rather advanced video and I realize that most people won't be able to recreate this either because this might be too intricate or because we use certain specialty tools here. We use Google Meets with the Gemini Assistant in there for the meeting transcripts. Then those files get stored in the Google drive automatically. We use air table as our database to store all of these meeting notes and make to glue all of this together for this entire process to happen automatically. And also I want to say a big thank you to make for sponsoring this video. I really like working with them because they just reached out and asked, "Hey, do you have any more detailed or advanced automations that you would like to show off? " And I said, "Yes, I actually have this one. " And they were all for that. Usually a lot of companies they want to do a extremely beginner friendly tutorial and Make does exactly the opposite which I really love. They say hey show something specific and show us something useful rather than showing a generic beginner tutorial. We've seen those and we want to go deeper and that's what we're going to do in this video. A final note before we get into the overview of this is that hey yeah if you're using a different tool stack and for example you might be using notion for the database and for example you're using a different meeting assistant to create your transcripts or you're using Dropbox instead of Google Drive. Well, you'll have to switch out those nodes and this tutorial just shows you what we have done and gives you that blueprint. If you want to customize it because you have a different tool stack. Well, this is a fantastic starting point. You can use the blueprints that we give you. Actually, just upload that to Chat GPT and then start talking to it, what steps you would need to take to customize this, but this tutorial is just for our workflow. So, with that being said, let's get into this. And I'll just start by giving you an overview. And then we'll get into the blueprint and how to

set this up for yourself. So, the trigger here is quite simple. It's a timed Google Drive connection that checks the meet recordings folder in my Google Drive. This is just where the Gemini meetings automatically go into. So, this is the tutorial automation, but the proper one is enabled to run every 15 minutes. We just have a few meetings a week. Nevertheless, this just costs one credit at a time, and we don't mind. Next up, there's a text parser and tools. And you'll see a bunch of these throughout the automation because not all of the formatting that comes from Google Gemini is ready to be worked with in some of the other tools. For example, we want to bring it from the meeting summary and the transcript within here into Air Table which stores dates beautifully like this. In order to do that, you need to parse out certain parts including the date format right here. For example, right here you can see what we do with the date format to make this fit into Air Table. Then we move on to actually downloading the files. And not just that, we also convert it to markdown. So we can easily make this work within this automation. Then we just make sure this markdown file is in UTF8 format. So all of this is well readable in the GPT API call that is coming up in a second here. Now we enter this router and this is basically here to filter out just the files that have everything we need. So if it's just the transcript, it goes here. But if there's everything we need, we go down this main branch down here, which is essentially the entire workflow. I mean, up until now, we kind of just prepared and downloaded our file. And now we're here in this main branch. And I'm going to show you this

on the example of a successful run on Monday. I tried to manually replicate it here for you with like duplicating a file and making it happen. But that made me realize that the way this automation is set up works perfectly when Gemini saves the file and puts everything in place. But for me to replicate it here, I would need to rename certain things and change parts of this automation. And I think that would cause more confusion than that's worth. So I'm going to show you this on the example of the automation that actually ran on Monday on our latest team meeting. So I'm just going to look at this run and this is what it looks like. Found the new file. went through all of these steps, downloaded it. Up here, it set one variable for the notes, which is the summary, and one for the raw transcription. And then it pulled in the recording here. It successfully went for this router as we have everything we need. And then back here, we get into using the GPT API. Let me briefly switch over to the editing mode so you can see this. It's all within the blueprint that we'll provide with this video. But as you can see right here, this is a prompt that we've worked out over several months. Here are the different blocks we wanted to include like the uncomfortable truths like identifying action items, listing longerterm high impact opportunities and more. You can check out all of this in here. Then we create this JSON that is going to be the structure that we'll be using for the GPT call. So the system prompt at the top and a transcription below it as context. And then here you set your API key and then you pull in this JSON to give the API all the info it needs to work properly. Hope that's making sense in terms of calling the API. That's pretty standard stuff. We've covered this dozens of times on the channel. Again, this is not exactly a beginner tutorial. Let's move on. And then here in the end, we do a bunch of formatting steps so that all of this data can be presented nicely within the air table. And then, as you can see, it takes the output of the last step here 58 to save that within the air table summary field. And if I look over to the air table, that's exactly what happens. It goes into here. And we can see the full custom summary in here. So, as I mentioned in the beginning, now we're not just getting the standard Google Gemini summary, which is, you know, fine, but it gives me this in-depth report here. So, there you go. That's a semi-detailed overview of what this actually does. And now I'm going to show you how to actually import and activate this for yourself because we exported this automation as a make blueprint, which you can in your free account even import. All you need to do is go into a new scenario, say import blueprint, and then get the file from the link in the description below. We'll put it up in a public G drive for you to download. So, I went through this process of setting this up myself a few days ago now. And really, there's three main things that you need to change. First of all, you need to switch this to your own connections. So, all the Google Drive notes, which are one, two, and then free right here, need to be changed to your own Google Drive. You're not going to have access to our company Google Drive obviously. And then you'll also need to adjust the meeting folder. Same thing goes for the Air Table connection. By the way, quick note is that you might have seen that in previous tutorials we used a lot of notion, but we actually switched to Air Table for all of our automations just because it's speedier. A lot of times we saw errors from various automations, especially with higher throughput and a lot of data coming through. And we needed some workarounds because Notion is freely available and it can work as a database, but if you put a lot of info into something like a long transcript, you would have to chunk it into different pieces and do all of these like data gymnastics just for it to work in Notion. In Air Table, it just works and it's speedier. The downside is that it's paid, but for those reasons, we switched everything over to air tableable and that's what we're using now. So, you'll have to reconnect that too and you'll need a air table that is set up like this. A name field that is a single text line, a date date text type, the Google doc link that is a URL, a recording link which is also a URL, and a summary which is a long text field. If you have this set up in Air Table, you'll add a new connection right here. And then just make sure that these exact variables link in the various fields and this will work. The same thing goes for this second air table node here. And then most importantly there's a third air tableable node here in the end. So if everything goes right, this is where it saves all of that. If you just copy the structure of this air table and you link your account, it should be able to transfer everything over smoothly from the blueprint. So you just need to switch out the connection, pick the right base and table and all of this should appear for you. Just in case you need to recreate it, I'll show it all like so. And then the last step is actually changing the value here in the OpenAI API call to your own API key. The rest can stay the same. And that should

make this full automation work. In case you run into troubles, you can always create a screenshot, give that to GPT or use some of the make documentation, which is very extensive to help you make some of these nodes work. Customizing this to other tools is obviously a different story, but I mentioned that in the beginning. And once you have it all working, here's my recommendation. You can customize this. Okay, so each organizations and each individual's goals are going to be different. And the point of this is that you can really go into here and play with things like the system prompt right here. And then at the end of this prompt is the part that is so crucial for you to customize yourself. So if you scroll down here in the preset, you will see this part that says mission vision values that specifies them and then the road map. Now you can delete parts of this. For example, now we're in Q4, so I won't need these parts. Just make sure to include what you're actually doing with your company and what your actual mission, vision, and values are for this to work properly. Otherwise, it's not going to know about you, right? So, I'm just going to copy over our mission from a file here on the other screen. Our vision is over here kind of if you're curious, I have like a notion knowledge base where I take these things from. But here, I'm just going to copy over the vision. And then finally the values that we kind of represent here and that I hold my team to. And then for the road map, I also have like a few road map points on things that we're working on right now. So I'm going to copy that over like so. This is just these are some items of our road map right here. And then when this is updated at the bottom, I'm going to make sure to save this. Now the automation knows this and every time it processes the transcript it will include my very own mission vision values and the road map from this system prompt note. You can try different ones. You can duplicate this automation and maybe have two versions of it with two databases where it runs two different prompts every time and then you can kind of split test and see what you like better based on your various meetings. Meetings are one of

the most powerful forms of context across the internet and there's something all of us do these days with remote work environments and Zoom calls being completely normal in most careers. All of those can be transcribed and with an automation like this you will be able to extract all of the knowledge from within that and it even allows you to do something like hey if you have an organization with let's say 30 employees and there's a bunch of meetings going on. Well this way you can be present in all of them or at the very least you can be scanning if everything that is being done in the meetings actually aligns with the road map that you set up. That's extremely powerful. Another example would be as chief marketing officer overseeing the entire marketing department and having a meeting analyzer like this for each marketing meeting that happens and then just getting these reports and seeing what uncomfortable truths are being avoided in the meetings. That's usually where the growth opportunities hide. I want to show you one more thing to round this out though. So this entire automation is happening in make but as I told you we're using air tableable as a database and what we did here within Air Table is set up a so-called interface. This is something that they have on their premium plans where it gives me this manager dashboard and once the automation is working, I don't have to bother with that anymore. That's the whole point, right? As an active operator in my own company, I actually want to go in here and interact with this easily. And this interface that we built here does that. So you can see the various content meetings show up here. And it doesn't just link to the Google doc and the recording. It shows me the entire custom summary. So right here, I can see all the uncomfortable truths from within here. And I have the same thing running for all our weekly team meetings, which is kind of a all hands meeting, whereas the content meeting is focused on the YouTube channel. And that's the power of this. If you set it up once, you can create these dashboards for executive functions within a company and give them a level of insight that would never be possible manually. It wouldn't be possible without LMS to analyze the transcripts. this digital work revolution where everybody's using Google Meets, Zoom, etc. to meet remotely. And it also wouldn't be possible with tools like make that connect all of these variables and produce results like this in the end. So

I hope this was helpful. You can find the blueprint in the description below. And again, a big thank you to Make for actually pushing us to create this video and to show you some of the advanced workflows that are actually working for us. All right, my name is Igor and I hope you have a wonderful