There are AI artists who generate images. And there are those who direct them.

Delia Mija belongs firmly in the second category. Based in Bucharest, she spent fifteen years as an art director in advertising before generative tools existed: building campaigns, reading briefs, making things that had to work. That discipline never left. It just found a new medium.

What she makes now looks like it could belong in the lookbook of a luxury house that doesn’t exist yet. Fashion imagery and video with the precision of a creative director and the freedom of someone who has learned to trust her own eye. Always a concept underneath. Always something slightly unexpected in the execution.

Her visual universe is immediately recognizable, not because it follows a trend, but because it carries a point of view. The kind that takes years to develop and can’t be prompted into existence.

In this interview, Delia talks about where she comes from, what feeds her work, and what she’d tell anyone serious about building a practice with AI.

Can you introduce yourself? Profession, nationality and what’s your background before AI? Does it still show in your work?

I’m Delia Mija, an art director in advertising, based in Bucharest. Romanian, YES, which I think matters more than it sounds, because growing up with a certain kind of Eastern European visual hunger, always looking West, always a little outside the center of things, gives you a specific relationship to beauty. You study it hard because you had to earn access to it.

Before AI I spent 15 years in advertising making campaigns, brands, the whole machine. I know how to read a brief, how to build a concept, how to make something that has to work, not just look good. That background is completely present in everything I make now. I don’t generate images, I direct them. There’s always a concept before there’s a prompt. The advertising brain never switched off, it just found a freer playground.

How long have you been using generative AI, and do you use it professionally?

Around 3 years, from the beginning, really, when the tools were still chaotic and inconsistent and people weren’t sure what they were for. I’m using it professionally every day now: client work, branded content, editorial, social. It’s not a side experiment anymore, it’s the practice.

Where do you find your inspiration: artists, movements, images, feelings? What keeps feeding your visual universe?

When you have tools that give you the ability to dream, inspiration starts coming from everywhere, and at first it becomes wild, almost overwhelming. With time you learn to tame it, to become selective.

For me it started with questions I was already asking myself. Can I really create a fashion piece from pure imagination? What does that brand concept I had 3 years ago actually look like? What if that object existed? I started looking at how the great designers think: Schiaparelli’s logic, Margiela relationship to craft. Not to copy, but to understand the grammar of a visual universe that has rules.

Then the references expand. Paolo Roversi’s light, Wes Andersen color way. And beyond that: architecture, ritual objects, the slightly wrong feeling of a fever dream. My cat. The light in my apartment at 4pm. Things that are already mine.

What keeps feeding it is the combination of feminine obsession and conceptual discipline. I’m drawn to beauty that has a logic to it, something underneath the surface.

Which AI tools do you use the most and do you have a secret sauce? Any settings, workflows, or habits others might not expect?

Midjourney, NanoBana, Flux, Kling, SeeDance etc. depending on what the project needs. Each tool has a specific intelligence, a specific failure mode. You learn what to ask of each one.

The secret sauce, if there is one, is that I treat prompting like copywriting. Every word is a decision. I’m not describing what I want to see, I’m constructing the conditions for it to appear. And I’m extremely specific about what I don’t want, which is just as important. I’ve also learned to recognize when a generation is telling me something better than what I asked for and to follow that.

But if advertising taught me discipline, this side taught me something else entirely: how to experiment, how to let the process take you somewhere you didn’t plan. In advertising there is always a mantra, a fixed idea you’re building toward. Here the most exciting moments are when I find something I didn’t know I was looking for. The exploration is the practice.

The habit that probably surprises people: I do a lot of deliberate nothing before I start. I sit with the concept. I know what I’m trying to make before I open any tool. That patience is the whole thing.

What were you trying to make before you found what you actually make?

I think I was trying to bring to life something that was already fully formed in my head: ideas about fashion and beauty that I could see clearly but couldn’t execute with the tools available to me at the time. The vision was always there. A very specific relationship to the feminine universe, to material, to how beauty can feel both constructed and inevitable. I had concepts that lived in my head for years, a certain kind of image, a certain atmosphere  and no real way to make them physical. AI didn’t give me new ideas. It gave my existing ideas somewhere to go.

Is there a moment in your process where the tool surprises you and what do you do with that?

Yes, constantly. But I want to be precise about what kind of surprise I’m talking about. The surprise that matters isn’t randomness. It’s when a generation finds a solution I didn’t consciously imagine but immediately recognizes as correct, like it reached into the concept and pulled out something I hadn’t articulated yet.

When that happens, I stop. I don’t override it, I follow it. I adjust the concept to include what the tool discovered. That dialogue is actually the most interesting part of the process: you’re not executing a plan, you’re having a conversation with something that has its own strange visual intelligence.

What I don’t do is mistake a happy accident for a direction. I’m still the one who decides what’s worth keeping.

Is there something you’re trying to say with your work, or something you’d rather not have to explain?

I’m building a world that reflects back a version of femininity I couldn’t find anywhere else. Whether that’s something I’m trying to say or something I’m trying to find I’m honestly not sure. Maybe both. Maybe neither. I’m exploring through my own lens, and I don’t know yet where it will take me. That’s exactly what keeps it alive.

What’s the first thing you’d unlearn if you started over and what would you tell someone starting today?

I’d unlearn the instinct to make things legible. I spent 15 years in advertising where everything has to communicate instantly: the brief, the concept, the image, the tagline. That efficiency is useful but it also makes you afraid of work that takes time to land. I’d give myself permission to be slower and stranger much earlier.

What I’d tell someone starting today: don’t begin with the tool. Begin with what you’re actually trying to make. The tools are extraordinary but they’re also seductive. They’ll generate beautiful things all day and you’ll mistake that for having a practice. The practice is having a vision before you open anything. The rest is just learning the language.

And: edit ruthlessly. A hundred generations, one image. That ratio is not a failure. It’s the work.

Explore her work on Instagram @delia_ai_creative.
On Behance: deliaAIcreative
On Linkedin: deliami