Stéphane Benini

AI Director / Artist

You each have more than twenty years in advertising and VFX. Secret Level, Mathematic, .monks: studios that don't mess around when it comes to quality. Tell us about the first project where you thought: “If we don't master this now, we're dead.” Not the first fun test, but the moment it tipped from playground to professional survival.

[Stéphane] For me, the real tipping point was the Black Eyed Peas teaser in 2024. At that point, the production clearly told me that Will.i.am was expecting my “touch” on the art direction of the project. Not a test. Not a playground. A very precise expectation, at the highest level. I found myself on an American team made up of talent from film, animation, and traditional directing. Extremely solid people, with a real culture of craft. There, AI was no longer a topic in itself. It was just one tool among others for holding a very high standard. If I didn't master that language at that moment, I'd become useless in the conversation. And with Bruno, as a duo, I'd say it was the NFL project in summer 2025. There, we were facing a huge machine, with industrial pressure, non-negotiable deadlines, and maximum visibility. AI had to fit into a very heavy production pipeline without breaking anything. At that stage, either you understand how to use it properly, or you slow everyone down. And that's not an option. [Bruno] Actually, that moment when technique and technology innovate and become a differentiator is everywhere in our line of work, and it was long before AI. During my first internships at Partizan, 30 years ago, there were a few young directors who were doing pretty well, like Gondry, Fincher, Koonen (when I say “pretty well,” I'm joking: those guys were already beasts!), and they were already tinkering, weaving After Effects into classic workflows, not as a gimmick but to improve their storytelling. From then on, if you're sensitive to that (and with After Effects in 1999, you needed a hell of a lot of sensitivity to grasp the thing's potential), you repeat the pattern with every new technological breakthrough, at least to give it a try. And even if not everything necessarily leads to a change in practices (3D TV, VR headsets…), you have to keep a playful side, an explorer's side, and get your hands dirty, just to see. With AI, having close friends who are pioneers like Stéphane or Gilles Guerraz, I watched the developments of the early years with curiosity, but also as someone who doubted the thing's real potential. It was while tinkering with Stéphane on a project to be made with Runway Gen-2 that I got it. I was able to generate in a few seconds animations that used to take me several hours in After Effects. Gen-2 also let me apply style matching to rough 3D scenes, and maybe that was the click: when I understood that the standalone gadget had become a tool that could fit into a complex production pipeline alongside other trades. Stéphane mentions the NFL film on which we directed the AI with Mathematic, and in the end it was just the evolution of my first AI + 3D hacks on Gen-2, only better and more efficient, but the logic was the same. In fact, I believe the interplay between AI and 3D has incredible potential.

Stéphane, three years is nothing and it's huge in this industry. What you did before calling yourself an AI ARTIST, could you still do it today without AI, or has that become unthinkable?

With or without AI, the essentials haven't changed since 2022: intention, emotion, and storytelling remain exactly the same. What I'm trying to say, what I want the viewer to feel, doesn't depend on a tool. AI hasn't changed my eye as a director. It has changed the speed and freedom with which I can express it. Before, the idea had to wait for the whole system to be ready. Today, the idea can exist right away. But the starting point remains the same: a clear intention, the right emotion, and a story that has something to say. Working without AI hasn't become unthinkable. But it would mean agreeing to go backward. And above all, it would mean giving up a creative freedom I had never really known before, except in documentary, but never in advertising.

Bruno, your background blends directing, production, VFX, and motion design. You're the guy who builds the pipelines. The first time you integrated a generative model into a classic production pipeline, what did it break, or what did it fix?

I dabble in everything to find the best tool for the project I have in mind, and it's the same with AI. Why use AI? It shouldn't be gratuitous. I have no problem making films without AI; I still direct dozens of “traditional” films and documentaries every year. AI has to let me fill gaps. The first times I used AI in classic productions, it met that need: a missing shot, faulty sound (AI in sound is also insane in terms of new possibilities), and also, I have to admit, to play around, in a “if I slip 1 or 2 AI shots into this classic production, who'll notice?” kind of way. For me, there are two paths with AI. The first is full AI: you're free to create worlds following your imagination, and it has incredible potential; it doesn't break anything, it creates a new field of creation. The other path is AI as a tool within a hybrid workflow, and that's incredible too: like 3D, like After Effects, it's a new multitool that lets you work better and faster. I really think it's all positive in terms of creation. You have to adapt to it, just as we had to adapt to digital, to 3D, etc… No choice. But our industries are full of creative people who are used to teaching themselves, testing, and evolving, and who will know how to work wonders with this new tool.

“Don't Press Generate” is the name of your workshop, but also your philosophy. What core idea do you want to pass on?

[Stéphane] “Don't Press Generate” isn't a rejection of AI. It's a reminder! A reminder that creation doesn't start with a click, but with an intention. The core idea is that AI moves very fast, sometimes far too fast. And if we're not careful, we confuse generation with creation. Pressing “generate” too early often means avoiding the real questions: why this story, why now, why this way. Our philosophy is to use AI as a tool for exploration, previsualization, and stress-testing ideas, not as a creative shortcut. Emotion, storytelling, and the eye remain human. AI comes in to amplify that, not replace it. “Don't press generate” is an invitation to slow down at the right moment in order to be more accurate, more precise, and ultimately stronger in what we tell. [Bruno] “Don't press Generate” is a headbutt to the jaw of AI slop, and it's our philosophy. We're sparing with what we share on social media because we're grinders: we craft our babies. Spitting out a lazy prompt to be the first to share some crap made with Seedance 8 or Kling 43/6 on LinkedIn isn't what we're about, and it does AI more harm than good. We encourage creative folks to do as we do: don't release anything unless it shuts people up with quality and storytelling. I have a hard time hanging out on LinkedIn, even though it's an incredible place for AI pros, because the amount of garbage thrown together with AI by clumsy hands makes my eyes bleed. With AI, you often have to accept throwing out your bad projects and your “too AI” shots rather than producing like crazy, but that's counterintuitive in the age of TikTok and mass consumption of junk content.