AI Video Production Company: How an Early Start Turned Into Delirio’s Biggest Edge

01

Starting when AI video could barely hold a shot

When Delirio AI began in 2024, generative video was a different medium. Clips lasted a few seconds, faces drifted from one frame to the next, camera language such as shot sizes and perspectives was frequently misread, and nothing guaranteed that the same prompt would give the same result twice. Making anything that looked like a film meant fighting the tool on every shot.

That is precisely why starting early was valuable. A studio that works under those conditions cannot rely on luck, so it has to invent ways of getting predictable results out of unpredictable models. Those habits did not become obsolete when the models improved; they became the foundation for using the new power well.

02

What the early years taught us: precision, order and ethics

Working with imperfect tools taught the team three lessons that still define how Delirio produces. They are also the three problems the studio set out to solve with its own research and proprietary software.

1. 1

   Bring precision to an imprecise tool

   Models struggle with the technical language of cinema and add randomness. Delirio chooses, shot by shot, between text-to-image, image-to-video and text-to-video, and builds references so the result is intentional rather than lucky.

2. 2

   Tame project chaos

   A single sequence can need hundreds of prompts, images and clips. Strict project management keeps AI “slop” from overwhelming a production and lets the team find, compare and reuse the right material.

3. 3

   Work ethically and transparently

   Every step of the creative process is documented, so no resource that breaks copyright or ethical standards enters the pipeline. Documenting early made compliance a habit rather than a late scramble.

The original toolkit combined video models such as Sora, Runway, Kling and Hailuo AI MiniMax with image tools such as Freepik Mystic, Midjourney, Flux and Google ImageFX. Learning when each one earns its place, and when to switch, is craft knowledge that only accumulates with hours of real production.

03

Why early experience compounds with today’s models

Today’s models are far more capable, but capability alone does not make a good commercial. A team that already knows how to lock a character, organise a canvas of references and document each decision can hand a stronger model a much better brief, and gets far more out of it than a newcomer starting from a blank page.

There is also a resilience effect. After hundreds of projects where the tool failed in new ways, the team stays calm when a shot breaks, because it has seen a version of the problem before. That is how deadlines are met without compromising quality.

The clearest recent example is the Wallapop campaign “Vendedores” with footballer Borja Iglesias, launched on 24 September 2026. Delirio generated the first 21 seconds with AI for the agency This is Odd, and live action takes over from second 22, a transition viewers do not notice.

A second recent piece shows the range: a Kodak Precise Next Lenses ad produced with AI for EssilorLuxottica, where brand fidelity and a polished commercial finish mattered as much as speed.

04

AI studio and AI VFX: what Delirio offers

Delirio AI is both a generative AI studio and an AI VFX studio: it creates images and video with artificial intelligence and finishes them with high-end post-production and visual effects. That combination is what lets it deliver broadcast-ready AI video production rather than loose clips.

The visual-effects arm is called Smart VFX and is delivered with Herokid Studio: a single solution covering product integration, advanced retouching and final artwork, always the smartest route to the result. It solves what generative AI cannot yet guarantee on its own: exact logos, legible labels and faithful brand colours.

For production companies and agencies, the studio works as an AI post-production and visual-effects unit that joins the team: it receives the brief, proposes the treatment and delivers masters and versions ready for TV, digital, social and outdoor.

05

The directors behind the work

Delirio’s directors come from film, music video and advertising. They bring narrative, framing, rhythm and direction of performance to AI, which is why a tool-savvy studio and a director’s eye reinforce each other. Watch each reel below.

• Albert Montero

  Barcelona-based director who moves between live action and AI production. He founded his own production company at 22 and has directed commercials for Ikea, Nike, Adidas, Bentley, L’Oréal, Ubisoft, Carrefour and Dunkin’ Donuts for agencies such as BETC, Publicis, Havas, TBWA, Ogilvy, FCB and DDB. His films have screened at Sitges and been recognised at the Epica Awards and in Runway’s AI film competition. Creative director at Sauvage.tv since 2024, where he has supervised AI work for Pepsi, Gatorade, Powerade and Under Armour. He directs performance as on a real set, so the result does not look like AI.

• Deka Brothers

  Ben and Julien Deka are French filmmakers based in Los Angeles. They won the 2012 MTV Video Music Award for Best Visual Effects for a Skrillex video, have directed for Gojira, Opeth and Carpenter Brut and for labels such as Atlantic Records, and have made advertising for JBL, Cox and StockX.

• Gami Orbegoso

  Director who blends traditional cinematography with generative innovation, under the motto “AI is the tool, but human passion is the engine.” Trained in acting at performing-arts schools in Barcelona and in independent cinema, with recognition at festivals such as Slamdance, he brings meticulous care for atmosphere, composition and visual storytelling.

• Ignasi AO

  Filmmaker and visual artist who mixes cinematic language, narrative and visual experimentation to create advertising, fashion and music pieces, exploring AI and visual effects with an author’s eye. Years of traditional filmmaking gave him a personal universe; AI let him make pieces with character: sometimes delicate, sometimes enigmatic and often brutal.

• Pablo Rivera

  Director working across AI and live action, with more than 15 years in commercials and branded storytelling before moving into AI-led filmmaking. His work blends cinematic craft with the visual language of contemporary culture, and is influenced by fashion, sport and hype culture, which helps brands stay current, distinctive and creatively ambitious.

06

An extra department for production companies and agencies

Delirio does not compete with traditional production companies; it complements them. As a specialised house for AI-generated clips, it slots into audiovisual projects like one more department, working side by side with directors, creatives and producers so the AI footage fits the rest of the production.

The studio also invests in research, experimentation and proprietary software, designed to make AI workflows more precise, efficient, organised and ethical. That is the long-term payoff of starting early: the tools Delirio builds today exist because it ran into the problems first.

07

Responsible by design: documented, licensed, certified

Because documentation began on day one, responsible practice is part of the workflow. Delirio uses commercially licensed tools only, keeps a record of how each piece was made, and signed the EU Code of Practice on transparency of AI-generated content, the voluntary instrument that helps providers meet Article 50 of the EU AI Act.

Pieces are also certified with VeriqX: the creation record is anchored with an eIDAS qualified electronic timestamp and a C2PA content credential is embedded in the file, so provenance travels with the work.