---
title: 'From Google Drive to Social Media: Fully Automated AI Workflow'
source: 'https://youtube.com/watch?v=N4Q4iM05PPc'
video_id: 'N4Q4iM05PPc'
date: 2026-08-12
duration_sec: 1426
---

# From Google Drive to Social Media: Fully Automated AI Workflow

> Source: [From Google Drive to Social Media: Fully Automated AI Workflow](https://youtube.com/watch?v=N4Q4iM05PPc)

## Summary

This video demonstrates how to automate the process of posting business photos to social media platforms using AI. The host and guest, Kevin, walk through three different automation tools—Make.com, n8n, and Claude Code—each offering varying levels of complexity and customization. The core workflow involves uploading images to Google Drive, using AI to generate captions, and posting them to platforms like LinkedIn and Instagram.

### Key Points

- **Automation Overview** [00:01] — The automation uploads business photos to Google Drive and automatically posts them to social media, saving hours per week by eliminating manual posting and hashtag writing.
- **Make.com Workflow** [01:10] — Kevin demonstrates a Make.com workflow where an image dropped into a Google Drive folder is analyzed by AI, a caption is generated, and the post is created on LinkedIn.
- **Setup Requirements** [02:45] — To set up, users need to connect Google Drive, select a folder to watch, and create an OpenAI API key via platform.openai.com, which is separate from a standard ChatGPT account.
- **Customizing Prompts** [05:43] — The AI prompt can be customized with business context, such as the business name, offer, and desired CTA, to generate more specific and effective social media captions.
- **Blo Dato Integration** [07:21] — Blo Dato is used to schedule and publish posts. Users connect their social media accounts and can choose the platform (e.g., Instagram, LinkedIn) for each post.
- **Scheduling and Limits** [10:17] — The automation can be scheduled to run at specific times, and the number of posts processed per run can be limited. Blo Dato also allows scheduling posts for future slots.
- **n8n Workflow** [13:17] — An alternative n8n workflow is shown, offering more flexibility and customizability. It uses similar steps but allows for parallel branches and uses Anthropic's Claude as the AI model.
- **n8n Advanced Features** [16:16] — n8n supports parallel execution of branches, which is useful for complex automations. It also allows for data pinning during testing and uses a merge node to combine outputs.
- **n8n Setup** [18:19] — Setting up n8n requires connecting Google Drive (with more steps for self-hosted versions), creating an Anthropic API key, and connecting Blo Dato with its API key.
- **Claude Code Interface** [20:34] — For the most advanced level, Claude Code is used to build a custom web application with a graphical interface where users can drag and drop images, generate captions, and post directly to LinkedIn.
- **Claude Code Customization** [22:10] — The Claude Code application can be customized to show specific buttons and actions, and it provides a visual way to approve posts before they are published.

### Conclusion

The video provides a comprehensive guide to automating social media posting, offering three levels of technical complexity. The key takeaway is that even non-technical users can start with Make.com, while more advanced users can leverage n8n or Claude Code for greater flexibility and control.

## Transcript

automation where you can just upload your business photos to Google Drive and have it automatically posted to social media. So, instead of having to go into Instagram, go into Facebook, go into LinkedIn, and manually choose your
manually write your hashtags, this automation will save you hours per week. automatically. All you have to do is drop your photos into Google Drive. You could not make it any simpler. And today I'm super excited to introduce you to
Kevin. He built this automation. He's going to walk you through it and super want to introduce yourself real quick? >> Hi everyone. My name is Kevin. I'm 36 living in Malta. I've been building AI automation solutions for several years
now. Started with Make, went into an 810. I will do a lot of work with code code, front end, back end stuff. So, yeah, super excited to be here and share these automations with you to help you move your business forward in a way that
I'm really excited for what you've cooked up today. Let's dive in. So, what happens here is we have Google Drive and we have make.com. And the idea is that we can take a screenshot. So, let's just do this on the fly. I always
tend to revert to dog training for some reason. So, are you okay if I go with that use case all the time. I don't know how it started, but somehow it just stuck. So, let's do our guy here. And the idea is that we have our image. We
just drop that into a Google Drive folder. So, let me delete the old one right here. Move to trash. And once we're in make.com, this and we can speak about this in a second here, but in my case I'm going to choose where to start.
showing you this live right now. And then, once I hit run once, this is going to watch the file in my folder, which is that dog training image. It's going to
download the file into make.com, make it shareable. It's going to analyze the image. So, in this case it figured out that there is a man eating outdoors holding up one hand, etc. It created a LinkedIn image caption for that. So, a
consistency, and those small witty little moments. This can be highly customizable based on your business. So, can also give some tips about that. Then we upload this to Bloデータ and then we create the LinkedIn post with the the
LinkedIn caption and the image itself. So, now if I go to LinkedIn, even happening?" It's a new business. Yeah. Yeah. Just by me uploading that image in my Google Drive, I now have this LinkedIn post
with that caption, with that image, etc. So, let me just delete this. And then, let's go back in here. So, I don't know if you have any questions or if you want to dive deep into some of this. I I want to mention some stuff
have? >> helpful to start with any setup required. So, for someone who's pretty new to make.com, what is the minimum they have to do to actually get this up and running, connect their Google Drive,
etc.? Which accounts do they need? >> So, first of all, obviously we need that once you're in here, you'd be able to hit add. You can give this a specific with your Google account. So, if you are signed in on your computer with a
specific Google account, you can just sign in with that just as if you're signing into that Google account. Once that is done, what you would need to do is select the folder to be watched. So, in my case I created the folder called
Sabrina. Then within it I created this automation one subfolder, which is what pretty much all that needs to be done in terms of Google Drive. So, any files that are dropped within that folder will then be watched since we've set that up
and the great thing is that once you've module, you can use the same connection for the second and the third. So, that's the Google Drive part. Then there is the Open AI part in this case. And I'm using
two separate modules here. There's the analyze images and then there is the create an output from a prompt essentially. Both use the same Open AI key. So, in this case, similar to how we set up the Google Drive connection, I
can go in here. I can click add. In this case, we're not signing into Open AI. API key. So, we'll go to platform.openai.com. viewers, if you don't already have a platform.openai
account, it's actually different from your normal ChatGPT account that you use every day. So, you do have to go here, platform.openai.com, your billing and stuff. But the good
news is once you do this once, you can use ChatGPT and other Open AI services >> Great point. And here you're only getting billed each time you make a request within make.com, for example. So, it's not a subscription like the
ChatGPT web version. So, once you're in here, platform.openai.com/api-keys, name right there, create secret key, and this will give you that API key right
here. So, you'd copy that, go back into Make, paste that in there, and now I have a Kevin two connection. So, if I wanted to have different AI Open AI there, right? So, let me go back. I believe it was the second one. And over
here, this is reading from the previous module's output. So, it's referencing that file so that it can use it. In this case, I'm using GPT-5.4 to analyze that image. So, the connection is established and that is how the image is interpreted
on that, giving the dog training image, that is how that output has appeared right here. >> And one cool thing here is you can load in a lot of your business information and context into the prompt. So, that
your business name?" Maybe you want a CTA like, "Hey, call us here at this number or visit our website to learn more." Like you can fully customize this text prompt. The version Kevin's showing here is just a really simple one just to
everything's working, but you can totally fully customize this so that it >> Yeah, yeah. To that to that point, I can say my offer is about providing the top say my offer is about providing the top 10 dog training abilities. I'm just
can say end the output with call us at whatever for more information. So, now I'm giving it more context in this system prompt in order to get it to give me a better LinkedIn prompt. So, if I were to run this again, and if we were
what I got the first time right here. Open up a text pad here so we can see what the before here was like compared to what we have after. So, if I were to run this again with a better system prompt, this should also be much more
specific towards my business now that it has some additional context. So, this is the Open AI section right here. So, this now gave me this output right here. >> Yeah, with a CTA at the end. Right? And
got to tweak the prompt, right? You can feed in any information about your business, etc. If you don't want hashtags, you can put that in the prompt. And so, really, really easy to customize this once you get the base
can see our program focuses on the top 10 dog training abilities, da da da, etc. And then there's the CTA at the end compared to this more generic version of that. So, the more specific you are in the text prompt system prompt right
here, the better the output would be. And obviously here we have the Bloデータ connection. And in terms of setting this up, all you need to do is choose Bloデータ. So, you you're adding the module there, Bloデータ. If it's your first time setting
that up, it's very simple and straightforward in the sense that you hit add, you go to your Bloデータ account, you go to settings, API, you copy your API key here and as that appears, this button right here. And then you just
Hit save, and that's how you establish the connection. Important point to mention is that over here, if I go to settings, accounts, I would obviously need to have logged in and connected my accounts
right here. And once I'm in there, I can just copy the account ID. I know you 100%. So, here >> Actually, I really liked how you described that really simply. Like I don't even think I need to do a separate
clip. Like that was it. Like just grab choose your account. Boom. I mean, I wish all third-party platforms were that straightforward. So, first version where it was like you had to do a hard-coded JSON request, but
it's it's come a long way. Yeah, 100%. So, anyway, here I can choose my platform and if I wanted to Instagram, LinkedIn, whatever the platform that I [music] want, I choose my account. What is important here and if you're
importing the Make blueprint, it's already been set up with these modules two pieces of information. There's the text, which is the result, aka what we got from Open AI, which would be that post right there. And then we have the
post right there. And then we have the media URL, which is from 11 right here, which is the URL for Bloデータ or the image that we uploaded. And yeah, this is what then creates the post, in my case, on LinkedIn, but it could be Instagram,
TikTok, whatever the social channel you >> real quick like if business owners want how would they do that? >> Yep. Let's go here. Let's just choose >> Yep. Let's go here. Let's just choose Instagram.
let's give this a shot. Oh, yeah. You don't need to post it, but yeah. Should So, Yeah, account. So, in my case here, since I'm connected to my Bloデータ just like we should with the API, I have the accounts that are available there. So, I
only have one account listed. If I go to my Instagram, so that is pulling directly from there. So, now all I need to do is give this the text version and then the media URL just like we did before. So, theoretically, So, you don't
actually even have to change the text or URLs. It's already there. And by the way, if you want to post on Instagram stories, you can do that as well. You just change the parameter that says post type from
post the story. And then, yeah, if you want to post to Facebook as well, you can just change that, too. Yep. And then, this is also pulling directly from can you talk about so how frequently
will this run? So, like how many photos should I be uploading per day? What if I going to happen? What should I be mindful of? Yep. So, over here, few variables that we can control. First of all, if we click onto that first module,
there is the limit right here. So, the maximum number of results to be worked with during one execution cycle. And if or when I turn this on, which I'll speak number that I've set here, this will grab X amount of posts depending on what
I've set here. So, if I had 10 images here, it would only grab two of them, right? Because I've said that to be two right here. So, you can either increase the number of posts that you would want to run per execution cycle. So, each
time this runs. The alternative, which I think is a better alternative, would be to play around with this scheduling every 15 minutes sort of value. Because if you click onto that every 15 minutes, this pop-up will come up and then you
can choose [music] when you want this automation to run. So, you can have it run every day at midnight, whatever the time might be, right? Every Monday, Wednesday, Friday at this time. And this would allow the automation to run and
grab whatever information is in that Google Drive. So, those are a few different parameters to play with. What you could also do, I guess we could we could get into scheduling here in a second as well. So, one feature in Blo
Dato is that you can schedule the next free slot. [music] So, if I go into my calendar here, I can set up the calendar and when I look at weekly schedule, based on what social media channels I've connected, I can add a slot which would
then connect to this right here. So, essentially, I can say, "Let's run this wanted to. I always know that when the time comes, twice a day, 10 times a day that would be going out. So, I can run this automation and post immediately.
So, I can perhaps that as option A, or I can also do this in bulk and then times of the week, there would be stuff that would be going out. Awesome. the automation before we wrap up? >> automation works well as is. You can
replace OpenAI with Claude or Anthropic if you prefer that. The other big thing, >> prompt right here. So, this is, I think, the biggest sort of leverage point that you can or you should invest time in when it comes to making this better. So,
if you have examples of what you like in terms of Instagram outputs, LinkedIn outputs, whatever you want that to be, this is where you give the AI exactly the things to not do. Awesome. Thank you so much for sharing. And if you're
and you're really looking to customize it or take it to the next level, Kevin's contact information is in this video if you want him to personally help you out. Okay, so that was the make.com version of this automation. And honestly, guys,
if you have zero technical experience, you're brand new to automations, I think But with N8N, you do get a bit more flexibility and customizability. So, Kevin's going to walk us through an N8N version of the same automation.
Pretty similar in terms of the flow from start to finish. First here, we have the search files and folders and then we're downloading the file. So, similar to make.com, what you might see somewhat different would be, first of all, if we
look at N8N's interface, if I zoom in here, there's the limit, which by default is 50. So, you can set this to two. Over here, I'm referencing the direct folder ID. By default, this would be from a list. You can click into that
and then you can choose your folder right here. I just find it to be very much easier if I just choose by ID, go to my folder and then just grab the ID right there. So, I'll just paste that in there and this is going to be all in the
same folder. Then, what I'm doing is downloading the file. Actually, let me just run this. We'll have some data right here. In the meantime, this has grabbed that file from our folder, right? The screenshot we took before.
And if I look at the next module right here, I have this download a file. So, I'm using the ID from this right here. So, if we look here, the first module or the first node in N8N found this image. The second one downloaded that data as
binary, which is just a fancy way of looking at data here in N8N. So, if I were to hit view, this would show me the visual representation of that binary data. Then, after I have that image or that binary downloaded in N8N, I'm doing
analyzing the image and then I'm creating the LinkedIn image caption in this case, same as we did with make.com. In this case, just for variety or to show what is possible, instead of using OpenAI right here or ChatGPT, I'm using
Anthropic's Claude as the model right here. So, I have the analyze image right here. And since I've downloaded the data, so here I have binary, here I chose binary file, then I'm giving it the input data field name, which must
So, data matches with data. And the output of that is the image description. So, the image shows a man kneeling down. After that, we have the create a LinkedIn image caption, same as we're saying before, the more detailed
will be. So, here I just have create a LinkedIn post caption based on this alongside the caption you're creating. Not the best prompt [music] by any means. Can be much, much better right here. But even that gave me this output.
So, excited to share a sneak peek. Yeah, this is from my run before. So, if I I need to run this again. So, one thing you can do in N8N is spin data, which is automations. So, instead of downloading
and running the stuff over and over, if you right click and choose spin or hit the P button on your keyboard, this will allow you to store that data as you are testing and building, which is why I got that historical information right here.
Since I ran this again with the dog training image, now I have something So, here I have, "Every great partnership is built on trust, that is the LinkedIn image caption. What I'm doing here is I'm using a merge
node. The reason I'm doing that is that when you download binary data in N8N, can get very complex if it's not output of the previous node. And here, since I'm using this binary data to analyze the image, I also want to upload binary
data into Blo Dato. Good thing is that for Blo Dato, you can either upload the binary data or I can turn this off and use a URL instead. So, that would then fields. In my case, since I had the binary information here, I just kept it
going with the binary information, which is why the analyze image and the upload media are connected to the output of this binary right here. Yeah, this is one of the neat things about N8N that you can actually have multiple branches
executing in parallel. You can't have that in make.com. And you may not need it for simpler automations, but as you get to customizing this and building out fine-tuned to your business, you may run into situations where having multiple
branches execute at the same time really handy for your automation. And that's one of the differences between N8N and make.com. And by just open up a run here, so don't run this again, what I'm doing here is I'm grabbing the output of
this branch. I'm grabbing the output of this upload media node, which gives me the URL of the image, and then I'm using this merge node to, as you can see here, So, combining everything together. So, I have the LinkedIn caption in this case,
I have the LinkedIn image URL through Blo Dato. So, then I can pass this to Blo Dato. So, then I can pass this to the create a post node in N8N, where I'm passing the text, I'm passing the image URL, and then this in return gives me
the LinkedIn post that was just made. So, on my LinkedIn, there's now that post that has been submitted. And if we go back to Blo Dato, publish posts, as you can see here, I have a few dog training posts right here. So, I can
also click into that and that will take me directly to this. And just to recap for the N8N automation, what is the minimum someone has to do to get started? So, what things can they kind of not touch and just leave it as is?
set up? >> Google Drive needs to be connected here. Now, with N8N, it depends whether you're on the cloud version. So, if I go to there's a cloud version right here,
easier. So, if you're on the cloud version, you would just go in here, would create a new credential, and then you'd essentially sign in to your Google account. If you're on the self-hosted version, like I am, then that would
involve a few additional steps. The great thing about N8N is that they have very good documentation. So, if I close this out, create a new connection or credential, and then hit open docs, there's a lot of information you can
set up that connection. So, you'd go to the Google Cloud Console, set up a new project, enable specific APIs right there, and it is pretty step-by-step in That's the first part in terms of the Google Drive connection. Then, there is
to connect to. So, in my case, I used Claude right here, pretty similar to OpenAI in the sense that I can create a new credential, and just like OpenAI, case, if I go to platform.co.com/settings/keys,
I can create a key right here, give it a name, hit add, and this will give me a key right here. So, copy that, go to N8N, paste it in here, delete me so I would remember, and then that is now ready to be used, right? So, turn it
connected to that, and I can now use that specific API key, connect to Sonnet this. And then, obviously, there is Blo Dato, just like we did with make.com. I can go in here, create new credential, give it the API key that again we got
from settings, API, and then copy API key. Again, this, once this is turned on and published, this will then run at that frequency. So, once a day at 10 minutes past midnight. One thing I really like here is this custom cron
expression, and you can be like very deliberate, very specific when this they give you here at this cron-job.guru, there's a bunch of examples right here, and you can also use Claude or ChatGPT
to really come up with something that is specific to you. This schedule trigger, enter the formula that you get right there. Okay, so now we've talked through make.com, and if you want something a bit more powerful, flexible, there's
n8n. If you want the next level, like Ferrari level when it comes to automation, then we have Claude Code. So, Kevin's going to show us real quick So, Kevin's going to show us real quick what you're able to do with Claude Code.
built together in order to really give you the ability to not just have the automation, but to have something in terms of a graphical user interface that once you start this on your computer, as you can see I'm running
you can drop an image here or click to browse. So, this right here, drag and drop the image, and then I can hit generate caption. So, on the back end this, based on the code that we created, it's following the or a very similar
you don't have to go into the automation and to an 8n or [music] make to get it to work. So, here if I zoom in, by clicking generate caption, this took the image, generated the caption right here, and this makes it very easy if you
wanted to edit it or if you wanted to change stuff based on the generated that's what we're all about. Let's say I wanted to just post that, right? So, I can delete that. Let's just, for example, I like hashtags, so I would
just leave that, and then I can hit post to LinkedIn. And this, already connected to LinkedIn. And this, already connected to posted directly from this custom web application that I just showed you here,
directly to my LinkedIn with my custom prompt or rather custom caption that was me editing the first draft of that caption that the Claude Code application generated. The great thing about this is that we're also giving you access to the
code behind it. So, once you download it, if you go in here, you can build on social channels out of the box. You [music] can give it your Bloop Data API key, the channel that you would want to post to, and this provides, I think a
nice visual interface for you to just drag and drop an image there, get it to just post to LinkedIn. >> Yeah, I really like that there's a visual way to approve all of the posts before they go out. That's one thing
that we didn't really show in the make and n8n automations. You can kind of do it with them as well, but it's a bit more complicated. Like you might write to a Google Sheet or an air table where you can see everything, but here, you
know, it's highly focused on the content task. You can see everything including your media and then one-click button to go ahead and post it. As Kevin said with Claude Code, you can customize this to whatever you need. You can tell Claude,
"Hey, I need this interface to show these buttons that do these things." And Claude will build it step by step. So, just to recap, guys, we covered Make, n8n, and Claude. Again, if you are a non-technical, I recommend just starting
with make.com. It's honestly the easiest one to get up and running. If you are comfortable with automations, you want that next level of flexibility, there's turbocharged level is you can truly build a custom interface and treat it as
system. Different employees can even access it. So, thank you so much, Kevin, for walking us through this. For folks who want more help figuring this out or can they reach you? >> Oh, I think LinkedIn would be the
easiest channel. Add me on LinkedIn, send me a message, happy to help out.