Some stories are impossible to film. Not because they are not worth telling, but because they are beyond the reach of traditional production.

Imagine placing a sports commentator overlooking Los Angeles. Or broadcasting from the middle of Times Square just minutes before kickoff. Or delivering pre-game analysis from a sandcastle on a private beach.

For Telemundo's World Cup coverage, those were not just creative thought experiments. They were the actual creative brief.

The network wanted to delight viewers with cinematic moments that would not be possible through traditional production. The challenge was to bring those ideas to life during one of the fastest-moving events in sport, where matchups change by the day, production timelines shrink to hours, and every broadcast has to meet television standards.

Working alongside leading entertainment and sports agency Creative Artists Agency (CAA) and Telemundo, HeyGen helped turn that vision into reality, creating four AI-powered broadcast segments that were aired throughout Telemundo's World Cup knockout coverage.

The project marked HeyGen's first major live broadcast partnership and showed that AI video had progressed from an experimental technology to a production tool ready for prime time.

This was not about replacing anything Telemundo had already planned. It was about adding something incremental and creative to a broadcast that was going to air in any case, using AI deliberately to deepen engagement and expand what the coverage could achieve.

Telling stories beyond production limits

The creative challenge was not finding interesting locations. It was making impossible locations feel believable enough to support the story.

Traditional production would have required multiple crews, international travel, location permits, visual effects, and weeks of post-production. None of those timelines matched a tournament where the next opponent would not be finalised until just days before broadcast.

Instead of asking where cameras could be placed, the team began asking a different question: Where should the story unfold if there were no production limits at all?

That shift opened up a completely new creative approach.

Designing moments rather than filming them

The final campaign showcased four Telemundo commentators placed within cinematic environments inspired by the tournament:

• Adriana Monsalve appeared overlooking Los Angeles.
• Pablo Mariño became "El Gran DT" in the middle of Times Square.
• José Luis López Salido delivered his analysis from a sandcastle on a private beach in Miami.
• Luis Omar Tapia stood alone at the centre of the pitch inside Dallas Stadium.

Each segment lasted 20 to 30 seconds and was aired during Telemundo's pre-match coverage from the quarter-finals through to the World Cup Final.

The objective was not to convince viewers that these moments had actually happened. It was to embrace AI as a creative medium capable of placing talent somewhere cameras could never go.

Rather than hiding the technology, the production highlighted and celebrated it.

Designed for broadcast, not just social media

Creating cinematic television is very different from creating a social media video.

Every segment had to comply with broadcast delivery requirements, sound natural to Spanish-speaking audiences, and allow for script revisions that often came in just a few hours before production.

The workflow brought together two specialised HeyGen teams.

The Avatar team transformed Telemundo's on-set captures into expressive digital performances using Avatar V, Voice N, and Video TTS, preserving each commentator's identity while enabling dynamic performances in completely new environments.

From there, HeyGen's production team created the cinematic worlds around each performance, compositing scenes, refining motion, upscaling footage, and preparing every deliverable for television broadcast.

The pace kept up with the tournament itself.

Individual generations took only five to ten minutes, allowing around twenty candidate performances every hour. Dialogue revisions could often be completed within fifteen minutes, giving producers the flexibility to respond to changing matchups without needing to rebuild entire sequences.

By the time the assets reached Telemundo, they were almost ready for broadcast, complete with ambient sound. The network's post-production team added licensed music and completed the final mix before going on air.

Built with talent at every step

None of this would have been possible without the commentators themselves. Each of the four Telemundo talents participated with full consent and retained control over how they appeared and what they said.

Working in partnership with the CAA Vault, CAA’s full-service solution for the capture, storage, and licensing of digital likenesses, voice, and intellectual property, the initiative placed talent in environments cameras could never reach, while enabling them to remain in control of their own name, image, and likeness throughout.

Perfecting the finer details viewers rarely notice

The technology made the project possible. The finer details made it truly believable.

Spanish pronunciation, pacing, and regional accents were refined through multiple review cycles with native Spanish-speaking producers. Each performance was evaluated for cadence, emotion, and authenticity before client review.

The production team also resolved numerous subtle performance challenges, from hand movements and facial expressions to voice quality and synchronisation.

Those refinements happened even as scripts kept evolving based on tournament results, squeezing production timelines into overnight iteration cycles.

The project was not merely an exercise in generating AI video; it was a full-fledged broadcast production built around AI.

A blueprint for the future of broadcasting

For decades, television production has been limited by logistics. If talent could not travel, if permits could not be secured, or if deadlines came up too quickly, certain creative ideas never made it on air.

This partnership pointed towards a different future. Instead of replacing cameras, crews, or traditional production, AI broadened what production teams could imagine to begin with.

Telemundo did not use AI just because it was new.

The network used it because it offered a new, innovative way to handle logistically challenging stories in terms of conventional production within the realities of live sports broadcasting.

For HeyGen, the project was more than just a successful campaign. It showed that one of the world’s largest broadcasters was prepared to rely on AI-generated talent during the world’s biggest sporting event, working to live deadlines, in a broadcast environment where quality is non-negotiable.

That milestone goes far beyond sports.

The same production workflow can help broadcasters respond to breaking news, enable entertainment studios to place talent in seemingly impossible environments, and give creators entirely new ways to tell stories without being limited by time, geography, or physical production.

The question is no longer whether AI belongs in broadcast. It is what the next generation of storytellers will create with it.