2026-06-07原文6 min readUpdated: 2026-06-07

Mythograph Atelier #1 - Abstract Art That Means Something to You

In this article, the author describes the inspiration behind Mythograph Atelier, an AI art studio that creates personalized abstract paintings. The idea combines a museum visit's impact, the vision of dynamic AI-native apps, and the concept of a curious AI that asks questions to understand the user before generating art.

SourceHugging Face Blog

Back to Articles

Team Article Published June 7, 2026

Upvote

Kasim Akpinar

kasimakpinar

build-small-hackathon

When I received the email about the Build Small Hackathon, I immediately liked the idea and wanted to participate.

In the registration form, there was an optional question asking what my project would be. I left it empty.

I did not have that amazing idea yet. I did not have the idea that felt like a serious competitor, or something obviously impressive from the first sentence. But I had a few “not bad” ideas, and I wanted to start building.

Most of those ideas came from two inspirations: a small but powerful AI agent skill called “grill me,” and my imagination of what apps might look like in the AI era.

Then a third inspiration came from a museum visit during a recent trip I had.

That is how I decided to start Mythograph Atelier.

(An example of the kind of abstract art that stayed in my mind after visiting IAACC Pablo Serrano. This painting is by Juana Francés, an artist I discovered during the visit and really liked. Image source: “Habitar la materia” - La gaRceta de la Ribera.)

Mythograph Atelier is an AI art studio that creates abstract paintings with meaning. The idea is simple: instead of generating a random beautiful image, the AI asks questions, tries to understand the user’s taste, emotions, ideas, and references, and then creates an abstract painting where the visual elements are connected to something personal.

The goal is not only to create an image that looks good.

The goal is to create an abstract painting that the user can look at and say: “This means something to me.”

You can try it here: Mythograph Atelier

Let me explain the inspirations.

Inspiration #1 - An Abstract Painting with a Meaning

A couple of weeks ago, I visited Zaragoza, Spain. Among the many interesting places in the city, one of the museums I visited was IAACC Pablo Serrano (Instituto Aragonés de Arte y Cultura Contemporáneos Pablo Serrano).

It is a contemporary art museum with sculptures, paintings, and many other contemporary art pieces.

Many of the abstract paintings in the museum had a beautiful use of color. I found them aesthetically pleasing. Some of them were easy to enjoy, even without fully understanding them.

But there was one painting that made me stop and ask questions.

It had shapes and strokes that almost looked like a child’s random lines. I looked at it and tried to understand whether there was a meaning behind it.

At first, I had a very ordinary reaction: “Can anyone just make random strokes and have it exhibited in a museum?”

Of course, I knew that could not be the whole story. There had to be context, intention, technique, history, or meaning behind it.

But I also felt something else.

I felt that it was a bit unfair when art speaks only to people who already know how to read it. If an artist creates something that only educated art enthusiasts can understand, what happens to ordinary people like me?

Are we just supposed to stand there and pretend?

But then I thought: maybe that is part of the beauty of art.

Maybe not everyone needs to find meaning in every piece. Maybe a painting does not have to explain itself to everyone. Maybe art becomes special exactly because sometimes you find a connection, and sometimes you do not.

And when you do find that connection, it feels like discovering a small treasure.

That became my first inspiration:

What if we could help people find abstract art that feels personally meaningful to them?

I really like the idea of an abstract painting with colorful shapes and lines, where I can see the meaning of Marcus Aurelius’ quote:

“You have power over your mind, not outside events. Realize this, and you will find strength.”

Not as text on the image.

Not as an obvious illustration.

But as an abstract composition that somehow carries that feeling.

Because the meaning, or the explanation, can be the connection between the piece of art and the person looking at it.

Inspiration #2 - The Future of Apps

When ChatGPT became widely known after its launch in November 2022, it felt amazing. But for a while, AI was still something I used from time to time: to ask questions, learn new things, or get help with small tasks.

My first interpretation of LLMs was simple: “This is like a very advanced search engine.”

After around a year, I had my real “a-ha” moment.

I watched an interview with Satya Nadella where he talked about agentic AI and the possibility of AI becoming the bridge between the user and the database, replacing many conventional SaaS workflows and interfaces.

That idea completely changed how I imagined the future of applications.

Since then, when I think about apps, I do not only think about fixed pages, fixed menus, fixed buttons, and fixed flows.

I think about dynamic interfaces.

Imagine an e-commerce website.

Today, most e-commerce websites still work with fixed categories, fixed product listing pages, fixed campaigns, and fixed offers. Personalization exists, of course, and has existed for years. But in many cases, the layout is still mostly the same for everyone.

The website may guess your preferences from cookies or past behavior, and then recommend products accordingly.

But the overall experience is still fixed.

AI can change this completely.

You could start with just a text box and write:

“I’d like to buy a budget-friendly pair of shoes.”

Or:

“I just want to look at flower pots and get inspired.”

Then the app could shape itself around your intention.

The categories, filters, explanations, product cards, comparisons, and even the visual layout could be created dynamically for that specific session.

This is my second inspiration for Mythograph Atelier.

I do not want the user to fill out a long static form.

I want the experience to feel more like a conversation, where the interface, the questions, the visual controls, and the final output are shaped by the user’s thoughts with the help of AI.

In the AI era, each app session can become unique. The UI does not have to be the same for every user or every visit. It can adapt as the conversation evolves, showing different inputs, options, examples, or visual elements depending on what the user is trying to express.

The experience can be as free as your thoughts.

Inspiration #3 - A Curious AI

A while ago, a colleague introduced me to an AI agent skill called grill me.

It was created by Matt Pocock, and it is very simple but extremely powerful. The skill is basically a set of instructions like this:

Interview me relentlessly about every aspect of this plan until we reach a shared understanding. Walk down each branch of the design tree, resolving dependencies between decisions one-by-one. For each question, provide your recommended answer.

Ask the questions one at a time.

If a question can be answered by exploring the codebase, explore the codebase instead.

I really like this idea.

Before starting a coding task, this kind of skill can be incredibly useful. It does not jump directly into implementation. It asks questions. It explores. It tries to understand the plan deeply before acting.

Of course, this cannot be copied directly into a user-facing app.

We cannot exhaust users by asking question after question forever. A good app should not feel like an interrogation.

But the core idea is powerful:

An AI should be curious.

It should make an effort to understand the user before producing the final result.

For Mythograph Atelier, this means the AI should not immediately generate an image after one prompt. It should ask a few thoughtful questions. It should understand the user’s taste, emotions, references, and the meaning they want to express.

But it should also keep the experience smooth.

That balance is important:

Curious, but not tiring.

Personal, but not complicated.

Creative, but still guided.

What Mythograph Atelier Tries to Do

Mythograph Atelier combines these three inspirations:

The desire to find abstract art with personal meaning.

The idea that AI-native apps can have dynamic, flexible experiences.

The power of a curious AI that listens, asks, and tries to understand before creating.

The current version is still early, but the direction is becoming clearer.

The user starts with an idea, a feeling, a quote, or a loose intention. The AI asks follow-up questions to understand the user better. Then it creates a prompt for an image generation model, with the goal of producing an abstract painting that is visually interesting and personally meaningful.

I want the final image to be something the user can explain to a friend.

Not like:

“I don’t know, AI made it.”

But more like:

“This mountain represents patience. The door is change. The empty space is uncertainty. The colors are calm because I wanted the painting to feel hopeful.”

That is the kind of output I am looking for.

Initial Results

I was unfortunately not among the first 1,000 registrants, so I missed the opportunity to receive the Codex credits. But I have an OpenAI subscription, and I decided to use Codex as my coding agent.

This helps me build faster, and it also makes the project eligible for the OpenAI prize track.

To avoid spending my limited credits too quickly, I did most of the planning with ChatGPT first. Then I introduced the plan to Codex and used it to help me build the application.

After a few back-and-forth iterations, I was able to generate the following image with FLUX:

The description behind the image was:

A mountain is not decoration; it is patient ambition. A door gives it a second force: a threshold into change. That is the idea behind the painting: I want something about trying to build meaning inside confusion.

This was one of the first moments where the project started to feel real to me.

It was not perfect, but it had something.

Further trials brought another image:

With this description:

A calm, geometric dance where lines suggest nature’s flow and empty spaces let you choose your path.

These images are still not exactly where I want the project to be. They are early, imperfect, and there is a lot to improve. But they gave me the feeling that the direction is right, and that Mythograph Atelier can become something interesting with more iteration.

After seeing this image, I felt that the project was starting to go somewhere interesting.

Not finished.

Not polished.

But interesting.

And that is exactly what I wanted from the first version.

What Comes Next

There are still many things I want to improve.

I want the conversation flow to feel smoother. I want the AI to ask better questions. I want the final prompt to be richer, more visual, and more connected to the user’s answers. I also want the generated paintings to feel less random and more intentional.

My goal during the hackathon is to keep improving Mythograph Atelier step by step, become eligible for as many prize tracks as possible, and try to collect all the badges.

But more importantly, I want to see whether this small idea can become something emotionally interesting:

An AI atelier where abstract art is not only beautiful, but personal.

Let’s see what the next improvements bring.

Stay tuned.

Spaces mentioned in this article 1