The AI Creator Working with Disney and CJ

This event was held with the support of Asan Nanum Foundation.

Anyone can make something with AI. Turning what you make into a business is a completely different skill, and what this evening kept coming back to was originality and story.

Putting Keep in front of his own name

The speaker was KeepKwan, an AI creator who's worked with Disney, the Korean Ministry of Foreign Affairs, the Korea Heritage Service and CJ. The nickname tells you a lot about him. He spent twelve years in business strategy at SM C&C drafting mid to long term plans, and first ran into AI in 2021. It fascinated him and it scared him. Was this thing going to swallow him? So he took his given name, Seokkwan Yoon, and put Keep in front of it. KeepKwan is a name he wears as a declaration to protect himself.

AI became his own arena while he was researching what his company should live on five and ten years out. Before pitching anyone he wanted to find out for himself what the technology actually was, so he did it on nights and weekends. He fed a classical Chinese novel he'd written in high school into an early beta of Midjourney. The output was crude, but seeing text come back as something with color shocked him, and he walked away convinced this would become a serious weapon in the content business.

Designing his own exit as a test

He didn't set out to quit. At first he hoped to build it as a new business inside the company. But the work turned out to be too much fun, and watching things take shape one by one, he tasted freedom. Even then he didn't walk out. He gave himself a year of tests instead, and only if he passed would he leave the company he'd served for twelve years.

The tests ran on two tracks, educator and artist. He'd studied business and lived in numbers and slide decks, but he saw himself as an artist, and he wanted to become an educator who teaches how to use AI and how to protect what's human with it. As an artist the bar was an exhibition under his own name. As an educator, an offline class with enough people signing up to fill it. He cleared each bar stage by stage. A strategist to the end, he ran his own resignation like a project plan.

I asked what's been best about his third and fourth year freelancing. Freedom. The worst part is also freedom, because with the frame gone he works weekends and dawns, and the freedom of time can quietly disappear on you.

Anyone can make. Selling is different

Connecting what an individual makes to a business is another matter entirely, and it's the same theme now echoing through startups. Don't just use AI, make money with it. His first collaboration was with Disney. He'd started his artist life on X, posting work steadily and exhibiting abroad until the opportunities followed.

He keeps asking himself why they found him, and the answer he lands on is story. A clear story, a clear concept, a clear worldview, held consistently across the work. He builds content around Korean tradition and mixes it with other elements, branding himself so that Korean tradition brings his name to mind. Companies and institutions look for AI artists precisely for that edge and originality. Which is why his classes always open with the same instruction. Before you touch AI, find the worldview and the originality inside you. Find your color.

Sketch first, then open the AI

Even his theme came out of ordinary life. On a business trip to New York he saw hanbok in a shop window labelled Chinese kimono. If I barely know my own tradition, he thought, how would a foreigner know it? So he decided to make content that teaches it. Then he found that AI didn't know hanbok either, generating kimono or Qing dynasty costumes instead. Fine, he decided. I'll teach the model and study alongside it. He still checks the historical accuracy of every piece.

So he sketches before he opens any tool. Not which model to use, but what's in his head, on paper first, and then the details. How the garment ties. Which traditional pattern is correct. He teaches the same way, and in a sixteen week university semester the first six weeks are paper and pencil only. Students think about prompts and outputs so hard that they forget to pull out what's already inside them, so he trains them to find subjects through writing, doodling, taking photos and walking through streets they already know.

Running himself like a company

He treats himself as a company and sets yearly KPIs. The long term strategy is to move beyond images and video into physical products. Last year the theme was wearing. From AI photo spreads to fabric to silk screen, step by step, until he produced a wearable hanbok based on the archaic letterforms of the Hunminjeongeum manuscript. In November he staged a runway of AI designed clothing in Gwanghwamun Square.

One piece photographed the dancheong patterns of Geunjeongjeon Hall at Gyeongbokgung Palace and swapped the traditional five colors for neon, so in warm weather the black garment gives up its neon patterns. This year the theme is scent, visualising fragrance rather than describing it, and next year is eating, with the ultimate goal a taste the world hasn't seen. He doesn't stop at AI output. He carries it offline, into products, into revenue.

He also sees job boundaries dissolving. A man with no technical background taught an AI creative class for the HR organization of a large company, and the HR staff, who turned out to be keenly interested in content, built HR materials as their final projects. Work that once meant asking the marketing team for help now happens inside the team. What he told the audience stuck with me. Everyone treats AI as a tool that generates things. For him AI generates time. When an eight hour task takes minutes, the remaining hours go to learning something new, and to keep the feel of his hands he studies pottery and bonsai.

Dealbreakers and the human one percent

The small group discussions circled two questions. What makes you instantly write off an AI output, and what's your own touch that puts human warmth back in? The dealbreakers converged across tables, and the most common answer was clumsy imitation of reality. When AI creates something that doesn't exist it reads as creation. An imitation of reality that fails falls into the uncanny valley. Consistency came up next. The first draft looks plausible, and by the second and third revision the face and the essence drift, so people go back to Photoshop.

Broken Korean fonts, or numbers shipped without a single check, also made the list. What surprised me was how many people felt that looking AI made is no longer a real problem on its own.

The human touch always came back to the person. One photo you took yourself, or one drawing by your own hand, creates trust. One participant deleted every skill downloaded from GitHub, built their own, and got their best results. Another trained a model on their own speech patterns to hold the human tone. One table put it as a number. Thirty percent of the effort still has to be human, in review and in storyline. Nothing ends with a single click. You grind through three thousand clicks.

Every table arrived somewhere similar. Return to yourself, and keep the sense that it's you using the AI everyone else is also using. As KeepKwan puts it, not a portfolio but a storyfolio, and what remains is the person building the story.

What companies can take away

Translated into working language, four things remain. Invest in originality rather than tools, because in an era when anyone can generate, companies go looking for a clear worldview and a consistent story. Train human thinking before AI, since sketching and fact checking pull out what's yours and raise both quality and distinctiveness. Connect digital output to physical, offline revenue instead of stopping at images and video. And treat AI as a generator of time, which means measuring what people actually do with the hours it hands back.

FAQ

Who is KeepKwan? An AI creator and educator, formerly twelve years in business strategy at SM C&C, who has collaborated with Disney, the Korean Ministry of Foreign Affairs, the Korea Heritage Service and CJ, building work around Korean tradition.

How does he keep historical accuracy with AI? He sketches on paper before opening any tool, researches details like garment ties and traditional patterns, and checks accuracy on every piece, effectively teaching the model as he studies.

What was the key business lesson? Companies hire AI artists for originality, not tool skills. A clear worldview and a consistent story are what turn AI output into commissions and revenue.

Join Bloom

Bloom builds offline rooms where people and technology meet. We run them in Seoul, and now beyond it.

• 💬 Discord community, where events are announced first
• ▶️ YouTube, full talks from past events
• 📸 Instagram, photos from our events
• 🎟️ Bloom app, upcoming events and Mixers