Upscale Day Paris: Joaquín Cuenca, CEO of Magnific, discusses ai, luxury, and the future of creative professions

At Upscale Day Paris, Joaquín Cuenca from Magnific analyzes the changes brought by ultra-high definition to luxury creative workflows.

This Wednesday, October 21, at the Salons Hoche, Upscale Day Paris brought together fashion, beauty, and luxury houses, studios, and creatives around the role of AI in creative work. Magnific presented Magnific One, a fine-tuned image model positioned as a qualitative answer to mass production. Its CEO, Joaquín Cuenca, looks back on the reorganization of his teams, the convergence between filmmaking and marketing, the demands of the luxury sector, and the role of the creative when facing increasingly integrated tools. Upscale Day Paris: What Was Announced Magnific One, a fine-tuned image model. According to figures presented on stage, it achieves a win rate of 68.5% against four reference models: Midjourney (52.2%), GPT 2 (46.8%), Nano Banana 2 (42.3%), and Seedream 5 Pro (39.6%). A major release presented by Martin Leblanc, CXO: single-frame generation, multi-frame, references, multishot, and editing. 2D and 3D driven by natural language. Integrated motion graphics, and a brand kit demo. Reaching the milestone of one million subscribers. An explicit positioning: Magnific is not a platform designed to generate social media content in a single click. During the panel "Elevating the Craft", questioned by the moderator, Joaquín Cuenca compared AI to a hammer: a tool that one must learn how to use. The Interview with Joaquin (CEO of Magnific)

I do a lot of work in schools, and after years of traditional techniques, the mindset shift is difficult—in companies too. Last year, there was a lot of talk about education. How do you feel the market is evolving?

What we were saying last year remains true: there will be more work for creatives, not less. But time is still needed to figure out which jobs stay and which ones emerge. For now, we are still doing much of the same thing as before, with new tools. We are seeing a new kind of production company emerge that didn't exist before. At our place too: we had to part ways with five people, but we are now recruiting audio and photo experts because we are making video, which we didn't do before. And the goal is not to produce a video in one click. If that's what you're looking for, this isn't the best platform. The work is about helping to make something truly good.

With the convergence of teams and the integration of image, video, 3D, and motion technologies, Magnific has transformed significantly. How did things go on the staffing side?

We have significantly strengthened the sales side: pre-sales, post-sales, people who do demos and support customers. Today, about 45 people are in this division. A portion comes from jobs that AI has eliminated for us, such as certain art direction roles or translation. We managed to reposition nearly 50 people toward sales, and others toward workflow and the studio. For those we could not reassign, we offered a clear severance plan, and we tried to find them positions in other companies within the group.

Luxury is a highly demanding industry: the crafted leather of a handbag, the weight and drape of fabrics, the stitching. This isn't AI's strong suit, and many rely on ComfyUI. How do you take it up a notch?

Luxury is very difficult. And we have to be transparent: sometimes, AI doesn't work. When we hit a roadblock, we have people who master the techniques to unblock it, and we succeed in about 90% of cases. If it doesn't work, it doesn't work: it's a tool with its limitations. It's like a camera: magnificent, but it only captures what's in front of it. Within these limits, there are incredible things. AI also opens up possibilities that weren't easy before: fantasy, animation. And the success rate is improving: 80% today, 90 then 99 tomorrow.

The production workflow is starting to stay within the tool, whereas before we used to head back to traditional software. MCP opens up batch processing and variations, which is what a lot of brands are looking for to feed their channels with a very precise art direction. You’ve been particularly responsive on these topics. Are you the one driving this?

It's a group culture. On motion graphics, for example, it was Martin himself who developed the first version of the product. To make a product that works, you have to talk to the customer and use the tool just like they do. Otherwise, you end up making something that comes close, that kind of works, but doesn't really work.

The market is structured around several hubs: Figma with Weave, Adobe with its legacy tools, where you already have plugins in Premiere and Photoshop, Canva with Affinity and Leonardo, and Magnific, a European player covering the entire value chain. You are also present in film production, an industry that is hard to move. What is your angle of attack?

I think filmmaking and marketing are going to converge more and more. Today, they are two worlds with completely different needs: when you talk to people in the film industry, they ask you for ProRes, specific technical standards. Marketing agencies have other expectations. But on the marketing side, we're going to make more and more short films. We're going to shift from "I'm making an image" to "I want to make a short film."

You’ve managed to establish a foothold where other platforms failed, including with major luxury houses. What's next?

We already have plugins that connect seamlessly to Photoshop, Illustrator, and even Blender. But I think we'll need to go elsewhere less and less. Before, you had to use After Effects for motion graphics, or Premiere or DaVinci for color grading. Tomorrow, it will be built right in: you say "I want that color," and it's done.

A developer recently recreated the equivalent of the Creative Cloud using AI. When you see this kind of news, what goes through your mind? That we could build a second Magnific tomorrow?

My answer won't please everyone. We are coming out of an era where writing code was hard and imagining the interface seemed easy. Copyright protects the code, not the interface, because it was deemed secondary. However, today, the