The Rise of AI-Native Media Companies: Inside the New One-Person Studios

Most articles about “AI and the creator economy” describe a general shift: tools get better, workflows get faster, one person can do more. That’s true, but it’s also abstract. The more interesting story is what happens when someone takes that shift all the way — not using AI to speed up an existing production process, but throwing out the process entirely and building a media company where generation is the pipeline, not a step in it. A handful of real operators have done exactly that in the last two years, at very different scales and in very different genres, and their specifics are more instructive than any general trend piece.

Neural Viz: one person, a television universe, no crew

Josh Kerrigan spent over a decade as a working filmmaker — film school, years of production work in Los Angeles, a TV pilot he’d sold that never got made. In early 2025 he went full-time on something he’d started as a side project: Neural Viz, a YouTube channel that produces “The Monoverse,” a coherent, ongoing science-fiction universe told entirely through AI-generated video, with a human writing every word of it.

The flagship show, “Unanswered Oddities,” is a mockumentary set after humanity has vanished, where alien creatures called glurons conduct Ancient Aliens-style speculation about the species that used to run the planet. It sounds like a gimmick until you look at the production process Kerrigan has described in interviews, including a Wired profile: he writes a full script with slug lines, action, dialogue, and camera blocking, exactly like a traditional shooting script. He storyboards each shot and generates key frames in Midjourney. Then — and this is the part that separates Neural Viz from generic AI slop — he performs every character’s lines himself in front of a webcam, and tools built on Runway’s Act-One map his real facial performance onto the alien characters, so the timing, the pauses, the specific choices of a trained actor survive the translation into a fully synthetic body. Voice work runs through AI cloning and synthesis tools in the ElevenLabs mold. A finished two-to-three-minute episode takes Kerrigan roughly twelve hours and about a hundred dollars in monthly software subscriptions — figures he’s given directly in interviews, not marketing copy.

The result isn’t a novelty account. Individual clips have racked up hundreds of thousands of views on YouTube and millions across TikTok and Instagram, and the channel has drawn enough attention that Hollywood production companies have reportedly approached Kerrigan about adapting the Monoverse for traditional television — which would be a strange inversion: an AI-native property optioned for a legacy medium. What makes Neural Viz useful as a case study isn’t that AI made a hit; it’s that Kerrigan kept every part of the pipeline that requires taste and craft — the script, the blocking, the performance — and only replaced the parts that used to require a crew: the camera operators, the creature-suit effects, the compositing team, the location.

Genre.ai: an ad studio built to skip the agency model entirely

The second pipeline redesign worth studying is happening in advertising, not entertainment. PJ Accetturo built his reputation with a viral ad for the prediction-market platform Kalshi that aired during the NBA Finals — a surreal spot with a cowboy grandpa and a beer-drinking alien, made in two to three days for roughly $2,000. For comparison, a broadcast-quality ad through a traditional agency typically runs weeks of pre-production and a budget in the hundreds of thousands. Accetturo used ChatGPT for scripting, Gemini to turn the script into shot lists, and Google’s Veo 3 model to generate the footage itself, and he’s been explicit that the creative strategy leans into “weird” concepts precisely because they cut through in a feed in a way a conventionally polished commercial doesn’t.

That single viral spot turned into Genre.ai, the studio Accetturo now runs as CEO, working with brands including Oracle, Popeyes, Qatar Airways, and David Beckham-backed wellness brand IM8. Genre.ai’s own account of its reach puts combined campaign views above 275 million, with more than 300 million people exposed to its ads in a matter of months — numbers achieved with production teams that are a fraction of the size of a normal ad agency’s, on timelines measured in days rather than the months a traditional creative-to-broadcast pipeline requires. The studio has reportedly turned down Super Bowl ad opportunities on the traditional model, betting instead that fast, cheap, distinctly “off” creative distributed natively on social platforms outperforms a single expensive broadcast moment.

The pattern: collapsing departments, not just speeding up tasks

What Neural Viz and Genre.ai have in common isn’t the genre of content — one is serialized fiction, the other is 30-second brand spots — it’s that both eliminated entire job categories rather than making the people in those roles faster. A traditional TV pilot needs a showrunner, a director, a director of photography, a production designer, a VFX house, and a post-production team, each handing off to the next. A traditional ad campaign needs a creative agency, a production company, a casting director, a location scout, and a media-buying team. In both of the case studies above, one person (or a very small team) holds the creative vision from script to finished frame, with generative models standing in for every department that used to require its own headcount, own schedule, and own budget line.

This is also visible at a larger scale, where funded studios are chasing the same structural idea rather than just the individual creators. Asteria, founded by two-time Oscar nominee Bryn Mooser, partnered with Moonvalley to build “Marey,” a video model trained only on licensed footage specifically to give traditional studios a legally cleaner entry point into generative production. Secret Level, founded by director Jason Zada with former Netflix and DreamWorks executive Christina Lee Storm as head of studio, is building an in-house tool called Liquid Engine explicitly to compress its own production pipeline for film, television, and games. Fable Studio’s Showrunner platform — backed by Amazon’s Alexa Fund — lets users generate entire custom TV episodes from text prompts, and demonstrated the concept with nine unauthorized AI-generated “South Park” episodes that drew more than eight million views before being pulled. None of these are one-person operations, but they’re chasing the same collapse of departments into a single generative pipeline that Kerrigan and Accetturo are already running solo.

What doesn’t show up in the highlight reel

It’s worth being honest about the limits. Neural Viz took Kerrigan a decade of traditional filmmaking craft to be able to write and block a script well enough that AI generation could execute it convincingly — the tools didn’t replace the skill, they replaced the crew required to realize it. Genre.ai’s model depends on brands being willing to run genuinely strange creative, which is a bet that won’t work for every category or every client, and “off-brand weirdness” is a strategy with a shelf life as more advertisers copy it and audiences get used to the aesthetic. Platforms including YouTube have also moved toward more visible AI-content labeling and, in some cases, algorithmic deprioritization of low-effort synthetic media, which raises the bar for anyone hoping to replicate this with less craft and less human judgment than these two operators put in. The one-person studio is real, but it isn’t a shortcut — it’s a different allocation of the same underlying skill, applied through a pipeline that no longer needs anyone else to execute it.