I want to stop wasting an hour choosing images. I made a visual GPT for note
When writing on note, what is the most 'irrational' part of the process?
For me, it is definitely everything related to 'images'.
The text is written. The structure isn't bad.
Yet, I spend time searching for a thumbnail, agonizing over illustrations, and repeating image generation, thinking, 'No, that's not it,' or 'This isn't right.'
Before I know it, an hour has passed.
And in the end, I run out of steam and just settle for the usual image.
I think this might be a 'bug' that eats up a creator's precious brain memory.
So, I made one.
A GPT that optimizes your entire note image strategy, from thumbnails to illustrations.
Its name is
note Visual Editorial Director (GPT)
It reads your text, decides on the direction for the thumbnail, suggests where to place illustrations, and even provides prompts for image generation if needed.
In other words, just by dropping in your article text, it takes over a significant portion of the decision-making regarding images.
If you upload your icon image, the tone won't be inconsistent across the entire article.
It feels like having a dedicated designer sitting right next to you.
If you're going to waste an hour choosing or generating images, I think it's much more human to spend that time brewing a delicious cup of coffee instead.
*Currently, this has moved to official sales.
Please check this article for details👇
Are you on the 'with illustrations' side or the 'without' side?
Until a little while ago, I honestly thought this:
'The important thing is the writing, so aren't images just noise?'
But in my last article for members, I tried adding illustrations, using the article by my creator friend 'ken' as a reference.
When I asked in the comments section, surprisingly, 100% were on the 'with illustrations' side.
I was the only one being stubborn.
I've reflected on that, and decided to start including illustrations from now on.
But if I, with my perfectionist tendencies, face each image one by one, I'll never be able to press the post button.
Fiddling with prompts and letting my life slip away is counterproductive.
That's where this "leave all images to the GPT" tool comes in.
As a test, I added illustrations generated by this GPT to a text-heavy past article I published.
From deciding on the title to creating five illustrations, it took less than 10 minutes.
Moreover, most of that time was just waiting for the AI to generate them.
In reality, it was finished before I knew it while I was scrolling through Threads or reading someone's note.
With this, I can keep going.
What can this GPT do?
Roughly speaking, there are three things this GPT can do.
Determine the direction of the thumbnail based on the note body
Suggest the placement and number of illustrations
Unify the tone of the entire article based on reference icons
To be a bit more specific, just by pasting the body text, it will suggest:
"For this article, it would be easier to read if you insert an illustration here"
"For this scene, this kind of composition fits"
"This direction seems good for the thumbnail"
It will suggest all of that.
It's not a perfect designer.
But it's overwhelmingly faster than thinking about images from scratch.
Above all, being able to delegate the task of "where to insert illustrations" is a huge help.
For someone like me who used to be exhausted by images every time before posting, it has become more than enough of a partner.
It's not perfect, but
Actually, this GPT still has a few minor bugs.
To be honest, it's only hitting about 80% in terms of perfectly tracing my thought process.
But I think handing over this imperfect process is my current way of doing things.
Rather than just lining up finished products neatly, I feel it's more like me to show the state of things as they are—
'This is as far as I've gotten,'
'This part is still a bit off,'
'But it's already good enough to use.'
I've decided to release it for free in advance to all of you 'accomplices' who enjoy the trial-and-error process along with me.
However, this is a limited release for one week only.
The URL will be available until Friday, May 8, 2026.
*The free release period for members has ended (2026.05.09)
If you join the membership and save the URL before then, you can continue to use it afterward.
I plan to keep wrestling with it behind the scenes until it becomes a perfect version of myself.
I'd be happy if you could help me grow it rather than just waiting for the finished product.
As a sample,
I tried adding illustrations generated by this GPT to a past article of mine that was 'packed with text'.
From deciding on the title to creating five illustrations, it took less than 10 minutes. Most of that time is just waiting for the AI to generate, so in reality, it's finished before I know it while I'm scrolling through Threads or reading someone's note.
...
The URL is released beyond this point.
I'm always waiting for you on the other side of the door.
note Visual GPT by KOTAYA
Usage Environment
At the moment, I recommend using the browser version.
It works on the smartphone app version too, but it can be unstable. Especially with the Windows app version, it sometimes deviates from the intended flow.
If it doesn't work well, please try the browser version first.
How to use
[Step 1: Paste the body text]
[Step 2: Paste reference images]
Upload an icon image or reference images. If you want to proceed without them, enter
"Leave it to you".
[Step 3: Decide on the style and title]
The GPT will provide three "recommended styles" and "SEO-conscious title suggestions" based on the reference images.
If you want to proceed to thumbnail creation as is, enter "Create thumbnail". If you want to proceed to illustration creation, enter
"Create illustration". If you want tag suggestions, enter "Tag suggestions". If you want to decide yourself, send the style number or title.
[Create a thumbnail]
"Create thumbnail".
A thumbnail will appear, so make adjustments as needed.
[Create illustrations]
"Create illustration".
A sheet with 5 illustrations arranged vertically will appear. If you want to make corrections, send the details, such as "The 3rd and 5th ones have the same pose, so please change them."
[Split the illustrations]
"Extract".
Press "Confirm", and in a few dozen seconds, 5 images will appear in .zip format. Click to download. If there is no response after 1 minute, please try "Ctrl+R" to reload the page.
A list of insertion positions will appear along with them, so follow that to insert them into your note article. If it doesn't appear, try entering "Tell me the insertion positions".
A suspicious dialog might appear, but please press "Open link" to download. Trust KOTAYA 🙏
[Create additional illustrations]
From v2.0, you can now create additional illustrations.
When the first 5 images are not enough, enter "Create another illustration".
[Suggest tag candidates]
From v2.0, it now supports tag candidates.
"Tag candidates"If you enter this, it will suggest tags that seem suitable for your article.
It provides suggestions while considering the article theme, words that easily reach readers, and angles that make it easier to be found within note.
You don't have to use all of them as they are. You can just choose the ones that fit your article from the candidates.
Troubleshooting
Sometimes, the GPT may head in a direction different from your intention.
▶Streaming interrupted
"Streaming interrupted. Waiting for the full message..."
An error like this appears, and no response is returned.
I think it's probably a temporary OpenAI failure.
"Ctrl+R"If you reload the page with , the response is often displayed.
▶Nothing happens with "Extraction"
There are times when communication is not possible due to a server-side failure. Please wait a while and try again.
▶A thumbnail was output during illustration generation
The AI may be dragged by the previous task content, and a thumbnail might appear. In such cases, try entering "Create an illustration".
How was it?
Honestly, I don't know yet if there is any demand for this.
However, I think anyone who writes on note has probably gotten stuck in the image-selection swamp at least once.
Getting tired while looking for a thumbnail.
Spending more time generating images than writing the actual text when trying to create illustrations.
Wanting to maintain a consistent look, but it ends up feeling scattered every time.
This is a GPT designed to reduce those little drains on your energy.
I'd appreciate it if you could try it out and let me know your thoughts in the comments or via DM.
How do you all usually create your illustrations?
Thank you for reading until the end.
I'd be happy if you could support me with a like or a high rating.
KOTAYA✿
