Freepik Spaces

If you are still copy pasting text and images between different AIs, you are wasting time and money because the pros are building automated workflows. And in this video, I'm going to show you how to do this in Freepik Spaces. Together, we will build a workflow from scratch that takes a product image and a description and automatically turns it into five viral Instagram ads together with a preview. Then I'll show you my advanced workflows that I also use personally. And I even share the link with you so you can copy them and use them yourself. One is a UGC video ad generator that uses the latest AI video models like Kling 3 and Sora. The other one is a YouTube thumbnail generator that creates three viral YouTube thumbnails at once following my best practices. Then it automatically analyzes and improves them and scales the images up to the perfect resolution. And all of this is contained in a single tool. You don't need many different subscriptions anymore and you have access to all the latest state-of-the-art AI models like ChatGPT and Claude Nano Banana 2 and all the different video and audio models. My name is Florian Walther and this is the AI tool corner where I review the latest AI software to find out which ones can actually improve our lives and businesses. So the tool is called Freepik Spaces. I will put a link to it into the video description below so that you can follow along. If you buy a premium subscription through my link, you also get a 20% discount. And I will put the links to my advanced workflows into the description as well so you can copy them. But don't skip the tutorial because otherwise you won't understand how to even use these workflows properly. So here in the Freepik dashboard, we have access to different AI tools like images and video generators. We will take a closer look at them later. For now, we want to go to spaces which are these canvases where we can build automations. And this is what we want to build first, an automation where we can insert any product and we get five Instagram ad ideas with a complete preview and a caption. This will be a lot of fun. Let's build this.

So we go to spaces and we click on new space. This is an empty space. Let's rename it to Instagram ad generator. Let's start by adding an assistant. We click here on assistant which adds the first note to our canvas. This is basically a chatbot. When we select it and press F on our keyboard, we automatically focus it on the screen. So just like ChatGPT, we can now talk to this. We could say, "Hey, how are you? Send the message." And we get a response back. But we don't want to use this like regular ChatGPT. We want to make this part of our automation. So we remove this prompt. And here down in the settings, we can add a custom prompt, which I've already prepared. Either type this out yourself or go to the link in the video description where you can find this workflow and copy the system prompts. So, this contains instructions for the AI. Analyze the provided product image and product info and then come up with a professional ad idea. The ad should look photo realistic and include a short catchy text. Return a highly detailed AI image generation prompt ready to plug into an AI image generator. Don't add anything that's not part of the image generation prompt. The image generator has the same product image provided for reference. So you can just refer to it by its name. So those are the instructions for this assistant. And later we will update this to generate multiple ads at once instead of just one. We can also rename this assistant. Let's call it image prompt generator. Makes it easier to understand what it does. Also, let's change the model down here from GPT-5 mini to Opus 4.5, which is a stronger model that generates better output. Okay. Next, I want to add my product image. I already have this on my computer. This could be any product. I simply drag and drop it onto my canvas here, which uploads the image file. There we go. Let's rename this to product image. And now we can connect this image to the image prompt generator. So here on the right side we have this image output. We can click on it, hold the mouse button down and connect it to the image input on the left side of the image prompt generator. Now we have connected these two and as you can see here the image is now an input to this AI assistant and it has access to it. Now we could also mention the image with @ directly in the prompt like this. But this is the manual approach. We want to automate this, right? So we connect it like this instead. I also want to allow the user to add some product information, not just the image. So we add another note. We can click on the little plus icon here. And this time we want to add a text which is just a text input in which we can put anything. I have prepared this product description. We can change the size here. So this just contains a few bullet points about this product and I change the name of this note to product info. We can also change the color of this note here with this button. make it yellow for example. Now we can connect the note the same way to this image prompt generator here to this text input. And now the generator has access to the image and the product description. Now of course the automation is not done yet but we can already run the assistant and see what output this generates. So this will now take the product image the info and the system prompt into account and it should generate an image generation prompt and there it is. We can maximize this up here with open preview. There's also a shortcut, the letter A on the keyboard. And this is the image prompt this assistant generated. The next step is of course to plug this in an AI image generator. And building these workflows is so much fun. I really love this. So again, we click on add note. And up here we also have different categories for organization. When we click on image, we have an image generator for example. We add this to our canvas. Put it over here. Let's focus it with the letter F and rename this to ad generator. When we hover over it, we have different settings that we can configure. We can select the image model. And I want to set this to Google Nano Banana 2, which is the strongest image model right now. Let's select this. Let's also set the thinking level right here to high to get the best results. We can select a style and a consistent character. We will do this later, but not for now. We keep the aspect ratio of one by one and the resolution 2K. And now what do we connect to our ad image generator? We need a product image itself. Of course, we can connect it here. And we need the image prompt which we connect as well. We don't need a product info because that's already contained in the image prompt, right? We could add a system prompt to the image generator, but it's not necessary because the prompt is already generated by the assistant. So now let's run this and see what we get. As a reminder, this over here is the prompt photorealistic advertisement featuring the Bullshark Testo supplement bottle positioned prominently in the foreground on a sleek black granite surface inside a high-end modern gym. Blah blah blah. Here we see a countdown until our image is finished. Let's see if this is what we actually get as the result. Yeah, this is exactly what was described in the prompt. And I think it looks super cool. And the bottle still looks exactly the same as in our input image here. And this is our first advertisement generated via an automation. But we can supercharge this. Instead of a single ad, wouldn't it be cooler if we generated let's say five different ads at once and then we can pick the one we like the most? This can be achieved with lists which is another note type which we can find here under utilities. And we have to learn how to use lists to really harness Freepik Spaces to its fullest potential. So let's add a list to the canvas. Let's rename it to image prompts. And now we go back to the assistant node and change the system prompt. I want to keep the same system prompt but change it to plural. So that instead of one ad, it creates five ad ideas. Then here on the bottom right where it says export as text, we can change it to export as list. And then instead of connecting the assistant to the image generator directly, we first connect it to the list like this. And then we connect the list to the image generator. And down here it now says X0 because the list is empty. And this would generate zero images. But when we now run the assistant, it will put five prompts into our list. So we select the image assistant and up here we can select run from here. When we now press this button, it runs the assistant again and all the notes after it. So the assistant finished, but this time it generated five different ad ideas and it exported them as a list. So now in this list here, we have five different entries and each contains an image generation prompt. We can also maximize this in a preview to see the full prompt. And now the image generator runs five times instead of just once. and it automatically creates another list afterwards which contains the generated ads. Let's wait until this has finished. While this is processing, let's rename this list to generated ads. That's the second list, the one that contains the results. And now with the small change, we get five unique ads every time we run this workflow. Isn't this cool?

We can even edit these images using AI via this button up here. We can insert an image prompt with the changes we want to make. Again, you can select the model down here or let it pick one automatically. But we want to add a few more connections to our workflow. We are not done yet. But first, I want to go to both lists again and I want to change this setting from keep items to replace items. This means the next time we run this workflow, it will automatically remove these five prompts and replace them for the next five prompts. And I want to do the same over here in the generated ads list. You still have access to these old generations. They are saved in your workspace. I just want to replace them inside this list to keep a better overview. Next, I want to add another assistant that automatically generates captions for each of these ads. So, we add another note again, an assistant, and we name this one captions generator. Let's put it here below the image generator. We set the model to Opus 4.5 again and we add a system prompt which says analyze the attached product image and product info and come up with a fitting caption for Instagram. Keep it short and catchy. Then we connect the generated ads list. This one here which has this images output. We connect it to the images input of the captions generator. And now it automatically changed the output to a list and we can't even change it. Why? Because the input is a list. It will insert these five ads one by one, run them through the captions generator, and it will automatically export another list with the five captions for these five ads. We will run this in a second, but we need one more input for the captions generator, which is the product info. That's all the way here on the left. We connect this to the text input. There we go. And if you lose overview over which node is connected to which other node, you can always click on it, which highlights the connections. So if I click on the captions generator, we can see we have a connection here and a connection here and the other ones are grayed out. Let's try this out by running just this captions generator note. This should now run five times and it automatically adds another list as output for the five captions it generates. So let's rename this list here to generated captions. And there we can already see them coming in. So now we have five ads and five fitting captions for each of them. Lastly, I want to generate an Instagram post preview. For this, you have already seen this earlier. I prepared this template. Again, you can find the link to this workflow in the video description. Let's drag and drop it onto our canvas here to upload it. And let's rename it to Instagram post template. And now, if you want, you can pause the video and try to build the final step yourself. So, we want to take these five ads and five captions and put them into this Instagram post template. How can we do that? The solution is coming now. So, we add another image generator because we need AI image generation to create this post template. Let's rename it to Instagram preview generator. Again, for the model, we select Nano Banana 2, which is the best one. This time, we also need a system prompt because the input is only the image and the captions, and we need to tell the AI what exactly to do with them. You are an Instagram post preview generator. Use the provided Instagram post template to generate a preview of the attached ad image plus prompt. You can change the size of the template or clip off the caption to make it fit. I want to keep this resolution as 2K, but change the aspect ratio to 4x5. And the input of this image generator is of course the post template. So we drag this over here. We need our five ads. So we connect them two in the same image input and our five captions which go into the text input. Now all of this is added to our Instagram preview image generator. And you can see that the number of generations automatically changed to five because we currently have five items in both our lists here. Now the interesting part is what happens when we have two lists as input. Does it add each image with each caption? So would we get 25 generated images or does it process them in the same order? And the answer is it's the latter. So it takes the first image and it combines it with the first caption. Second image with the second caption and we get five outputs. And this is exactly what we want. Now let's run the preview generator because we already have our ads and captions. We don't need to regenerate them. We only want to run this node here. And off it goes. It generates another list. Let's rename this to a generated Instagram previews. And it here it will put five new images. And there are our results coming in. So it always connects this ad to the first caption. Second ad to the second caption. So this one says unleash the beast in the image and the caption starts with dominate your workout and swim with the sharks. Let's take a look at the result. And indeed unleash the beast. Dominate your workouts and swim with the sharks. Now, the generated image has some glitches in here, but we can regenerate them if we are not happy. The other ones actually look better. This is a perfect Instagram preview with our ad and a caption. And I think this is so cool. If you're not happy with a result, regenerate it or use the editing tools to change the parts you don't like. And then when you are done, you can download the image here and use it for your marketing. And you can now rerun this workflow with any product. and you always get five high-quality Instagram ads. You can replace the product image here by clicking on this replace button. Don't delete the note because if you delete a note, the connections are also gone and you have to reconnect it manually. It's better if you click on the replace button because this will keep the connections active. And then just upload your new product image. Insert the product info here on this note. And then go to the image prompt generator and click on run from here to run this whole workflow all the way to the end. Now all your past generations are stored in your workspace. So you can always find them. But what you can also do is you can select everything here in this current workflow and you can create a group. You can name this group like the product Bullshark Testo and then you can select the group. Press Ctrl D to duplicate it. This way you can keep all your products in this workspace if you prefer. Rename this, use a different product and generate the ads here. You can give the group a different color. How you use this is up to you.

We also have many more different notes available. We will use some of them in a moment like the video generators. We have music generators. You can look through this and we can also leave comments on the workspace. If you want to work on this with your colleagues, you can leave comments here and your co-workers can see them. Next, let's take a look at my UGC ad generator. Again, the links to all my workflows are in the video description below. So, this one takes three different inputs. A product image, product info, and an AI influencer. I have trained the AI on this fake person. Those are AI generated images. As you can see, I uploaded different reference images of her so that the AI creates a consistent character throughout all these different ads. I called her Era and you can upload and train your own character. I also trained the AI on myself, which is what I use in my thumbnail workflow that we will look at next. But for my UGC ads, I use her because she looks better than me. Then we have a viral video concept generator. The purpose of this is to only create the ad concept. You can see the output includes timestamps and here at the very beginning I have these instructions. My viral UGC formula. This contains best practices on these UGC ads. They are automatically inserted into this AI assistant that generates the video idea. Then we have another AI assistant that takes this video idea and it turns it into a generation prompt. Then we have a starting prompt generator which only creates the starting prompt for this video based on the video prompt. And then an image generator right here which creates the starting image. This uses Google's Nano Banana. Aspect ratio is 9 by 16 for social media. And this is just an image. And then we have three different video generators that take this start image and turn them into videos. Now this can get expensive because these video generators are expensive. So it's better to just pick one of these models and generate a single video instead of three different ones like I did here. I only did this for presentational purposes. This one uses Kling 3.0. Sora and then down here Grok. But we have a ton of other video models that you can pick from here under models. Just look at this huge list. They are all available right here. They all have different strengths and weaknesses. Let's take a look at the result. This is clean. >> Your friends say you're too old for Pokémon merch. Meanwhile, I'm freezing in my apartment every winter. And this just changed everything. Feel how soft this is. Official Pokémon. Crazy warm. And I literally live in it now. Links in bio and you need this. >> I think this is amazing. The other ones are not as good. >> Your friends say you're too old for Pokemon merch. >> Because these models are not as good as Kling. >> Meanwhile, I'm freezing in my apartment every winter and this just changed everything. Feel how soft this is. >> This output has some glitches. Let's take a look at Grok as well. >> Your friends say you're too old for Pokemon merch. Meanwhile, I'm freezing in my apartment every winter. And this just changed everything. Feel how soft this is. Official Pokémon. Crazy warm and I literally live in it now. Links in bio. You need this. >> Yeah, Kling is definitely the best one currently. And up here we even have video editing features. And here we can modify this video either via a prompt. You can tell it what you want changed. Again, you can pick from different models. You can add special effects right here. You can change the color grading. You can change the voice sound effects. trim the video and so on. And if you want to create the longer videos, that's also possible because all of these video generators have the end image as output. That's the last frame. So you could use this last frame, put it into another video generator as the first frame. That's this input here, start image. So you can connect the last frame to the first frame, add another prompt, and generate the second part of the video. And then you can use another video node called video combiner to combine these different clips into one long video. And not only that, you could also use one of the music generators. Select a music model like ElevenLabs or Google Lyria and add background music to your video. Consider this your homework. You already know enough about Freepik Spaces to build this yourself.

Next, let's look at my YouTube thumbnail and title generator, which I use all the time now. So similar to the UGC generator, this has a document with all the best practices which are contained here in this note. They are added to a concept generator. Again, the input is the video transcript. There are websites that allow you to extract the transcript from a YouTube video link. I will put it into the video description as well. Then you insert the video transcript here. Some additional notes if you have certain preferences. And in the assets is me. That's the character I trained. again using different images of myself so the AI knows how I look and we can attach additional assets like a screenshot that I want to have in the thumbnail and this workflow is pretty advanced. So first it generates thumbnail concepts but not one three at once by using a list. So this list now contains three different thumbnail concepts which are then inserted into an image prompt generator. So this takes these three thumbnail concepts, turns them into highly detailed image prompts which again are put into a list. So now we have three image prompts which are put into an image generator that generates the thumbnail. Besides the prompt, this generator also gets the assets me and the screenshot and a system prompt with some instructions right here. Since we insert three image prompts, we get three results, right? Which are put into this list. So now we have three thumbnails. But this first iteration is often not good enough yet. So I have another AI assistant which is my thumbnail optimizer. So the purpose of this assistant is to first rate my thumbnail based on the best practices that we had at the very beginning. So they are put inside this AI assistant which rates these three thumbnails on a scale of 1 to 10. And then it comes up with improvements based on my best practices that we can do on these first versions to make them even better. Again, these improvements are added to a list, turned into an image editing prompt, which are then added to another image generator. But the purpose of this one is to not generate a completely new thumbnail. Instead, it takes the three existing thumbnails. It takes the improvement prompts and it edits these thumbnails based on the improvement prompt. I hope you can still follow along. Meaning that this is the first version here of our thumbnails. Let's look at this one for example, the third one which is a bit too busy, right? There's too much happening in here. And then after our improvement step, this is the final version. As you can see, it's simpler, less busy, and it just looks better. So, this is the final thumbnail. But often the dimensions are not quite correct or the resolution is too low. So I have more steps, a thumbnail cropper which turns them to exactly 16x9 aspect ratio and an upscaler which uses another AI to upscale the image to a higher resolution. And here is the final result. So the thumbnail is big, high quality and it has the perfect aspect ratio. And the image you can see in the background is the asset I attached at the very beginning, this screenshot, because I wanted this to be part of the thumbnail and the AI did exactly that. Now there's one more step. I also have title best practices and another AI which analyzes the title best practices, the generated thumbnails. They are going from this note which contains the three thumbnails into the title generator best practices and also the full transcript here at the beginning. All of this goes into our title generator which generates five viral title ideas for each of these three thumbnails. So now in the results we have our thumbnails right here and we have the perfect title ideas which we can expand. So this first analyzes the thumbnail and then generates the best titles for it. And it does this for all three thumbnails. This is a bit more complex but you don't have to understand all these steps in the middle. You only have to insert these inputs. Click start and it will do all of these steps automatically. And at the end you get these results, the thumbnails and the titles. It's that simple to use. You can copy this workflow and run it yourself to make your own thumbnail. And in the past, I would have to do all of this manually. I would open ChatGPT, brainstorm an idea, then ask it to generate an image prompt, copy the image prompt into Google AI Studio, then again copy it into ChatGPT, ask it to come up with improvements, edit it, and so on. Now, all of this happens with only a few button clicks. And because this doesn't use video generators, it's also pretty cheap to run this because image generators are not as expensive. And on YouTube, you can then run AB tests to try these different thumbnails and titles. Now, as I mentioned earlier, in these image generators, you can also pass a style. So, if you have a certain thumbnail style that you want to follow, you can insert it here. I don't use this because I'm fine with these thumbnails looking different each time as long as my face is in them. But if you want, you can define a style here. And this makes this a fully blown AI thumbnail generator.

But if you want to see more exciting use cases, keep watching the video. Here under templates, we have a ton of more examples of what you can build using Freepik Spaces. And this is great to get inspiration and see what's possible. Let's take a look at one or two of them. For example, this multiformat ad production. So here you insert a product image similar to what we built earlier and a logo and it creates product images and ads in different designs. Looks pretty cool. And then it has additional steps where it generates different variations. This really shows you what's possible with Freepik Spaces. You can do anything basically and this looks super high quality. If you want to use this workflow, you can click here on use template which copies it into your own workspace and then you can modify and use it. Another one that I found pretty cool was this professional headshot generator. So here the input is a photo of yourself. This can be very casual. Then we have different instructions and prompts and it automatically generates high quality head shot in different scenarios. Again, this uses lists like we did earlier and you get professional headshots. And again, this replaces other dedicated AIs for generating headshots. I'll give you my final summary and who should use Freepik Spaces in just a moment. But first, let's take a look at these other AI tools we have available. We have a regular image and video generator where you can select a model similar to how we did it in our space. You can select a character, give it a prompt, and you can use all these AIs in here directly. And again, you have access to all the state-of-the-art models. All these different models are available in here. You can even filter them by features. And down here under all tools, we can find even more because Freepik is a huge app. Let's also take a look at the pricing and who should buy which subscription because these features are not free unfortunately. Now again, if you buy a subscription through my link in the video description, you get a 20% discount on premium subscriptions or higher. First, it's important to mention that all these paid tiers include a commercial license. So, you can actually use these AI generations for commercial purposes, which is important. Now, what's interesting is that the premium plus subscription gives you unlimited use on some of these models like Nano Banana 2. Whereas on the regular premium tier, you have to pay credits for each use. Here, they are unlimited. However, if you mostly generate images, then the regular premium tier is enough because you get 240,000 credits a year, which is enough to generate many images every day. However, if you were generating videos regularly, then you need more credits and you probably need a premium plus subscription. Premium Plus also allows you to top up your credits, so purchasing one-time credits, which the lower tiers don't allow. And of course, since Freepik was originally a stock image website, you get access to over 200 million premium assets that you can just use right away. This includes photos, videos, graphics, even fonts and you can use all of them in your own media like your YouTube channel or advertisements. So, what's my final summary and who should use Freepik Spaces? So the cool thing is that you can build any AI workflow in Freepik Spaces. No matter if it includes videos, images, text, even audio, you can build it all directly in here. This replaces a ton of other tools which would require separate subscriptions. But of course, it's quite complex. If you look at my thumbnail workflow, I actually built this by hand step by step with a lot of trial and error. We have to make the connections ourselves. We have to give the correct system prompts. So, if you're very technical and you have some patience and you're willing to build this yourself, then Freepik Spaces is an amazing tool for you. Or, of course, you can use one of the predefined templates and just modify them. But if this is too complicated for you, there are also dedicated AI tools that do specific tasks without you having to build a workflow yourself. For example, thumbnail generation or UGC influencer ads. This requires a separate subscription, but it makes it easier and often the results are also better than what you generate yourself using your own custom workflow. And I have videos on many of these tools on my channel. Check them out. So, you have to ask yourself, do you want everything in one tool with full control but more complicated setup or do you prefer ease of use but have to pay a bit more? But if you are a designer, Freepik is very useful anyway because you also have access to all these different stock assets. Again, all the links including my workflows are in the video description below. You can copy them into your own project and either use them as they are or build upon them. And if you like AI automation tools, I have a full playlist on them, which I will put here into the end card. Check them out next. Then I hope I see you in the next video. Have a nice day. Take care.