In this episode, we interviewed Elettra Fiumi - - AI filmmaker and founder of Fiumi Studios, a Lugano-based production company making documentaries, branded films and AI-native content for hospital groups, corporations, foundations and universities
Elettra brings more than fifteen years of international documentary and branded work - - with credits across Netflix, Amazon, the BBC and The New Yorker, and festival premieres at DOC NYC and Locarno - - now rebuilt around custom AI pipelines
Her core argument: the scarce skill in AI filmmaking isn't generating footage, it's the judgment to know which option to keep - - and the boring, repeatable pipeline a paying client will actually trust
Key takeaways:
The unglamorous problem is the valuable one - - everyone chased viral one-shot demos while the real need was a repeatable pipeline a client trusts across ten episodes and three rounds of revisions
Regeneration replaces the reshoot - - when legal wants to change a line in episode seven, she regenerates the shot in an afternoon instead of recalling a crew
Access is not capability - - a model spits out a thousand options in a minute, and the whole job is knowing which one to keep and which nine hundred and ninety-nine to kill
AI production becomes a normal budget line - - within twelve months mid-size brands request AI-native as the default, because the math on ten episodes versus one shoot is impossible to ignore
🔗 Connect with Elettra
Who are you and what do you do?
I'm Elettra Fiumi, an AI filmmaker and the founder of Fiumi Studios, a production company in Lugano that makes documentaries, branded films, AI-generated content and hybrid solutions for clients like hospital groups, corporations, foundations, and universities. I build the production pipelines that let a small team ship films that used to need a crew of thirty and a six-figure budget.
What problem did you see that everyone else was missing?
Everyone was struggling with no budgets, then arguing in fear about whether AI would replace filmmakers. Meanwhile the demos going viral were generic, one gorgeous shot with no story, no client, and no delivery date. I came from fifteen years of documentary and branded work, so I knew what clients actually need: a strong story, meaning, emotion, clarity, consistency, revisions, a character who looks identical in episode one and episode ten. The problem nobody wanted was the boring one: building a repeatable pipeline a paying client will trust across ten episodes and three rounds of revisions. That unglamorous reliability is what most of the field skipped straight past.
The problem nobody wanted was the boring one: building a repeatable pipeline a paying client will trust across ten episodes and three rounds of revisions.
Walk us through one concrete way your work changes what companies actually ship
Take the compliance training series I'm delivering for a corporate client: a package of films to train employees. The old way means hiring actors, booking a studio, shooting for days, and you still can't change a line without a reshoot. My way: we interview and research, we write the scripts, generate each episode with a consistent cast using a set of AI video tools, then package the whole thing with a built-in quiz that plugs straight into their learning system. When legal wants to change a sentence in episode seven, I regenerate that shot or voice in an afternoon instead of recalling a crew. Same budget bracket as one corporate video, various episodes, and a format the client can update forever.
What's the most common thing senior leaders get wrong about AI?
They treat AI like a vending button: press it, a finished film drops out. So they buy every tool, hand the team logins, and wonder why the output looks like everyone else's. Access is not capability. A model generates a thousand options in a minute, and the whole job is knowing which one to keep and which nine hundred and ninety-nine to kill. The scarce skill now is judgment, which largely comes from experience. The leaders who get this hire for taste and treat the tools as brushes, never as the painter.
When legal wants to change a sentence in episode seven, I regenerate that shot or voice in an afternoon instead of recalling a crew.
What's in your AI stack? The one tool you rely on every week?
My stack runs deep. I'm an AI Cinema educator and I'm a Creative Partner with most of them: Seedance (Dreamina), Figma Weave and Runway for video, ElevenLabs for voice. NotebookLM is always in the mix, too. But the tool I rely on every single hour is Claude. I've built countless custom skills and a Claude Code setup that runs the studio, from writing client proposals in my own brand style to generating quiz packages and call sheets. It holds the operational spine of the business so my hours go to the creative calls and story development. The generative tools make the images. Claude makes the studio run.
What does your work actually look like day to day?
The headline is "AI filmmaker." The real version is running five to ten client projects in parallel from a studio in Switzerland. Right now that's content for a hospital group, a recruitment film for a university, a compliance series, an exhibition, an experimental short film, a series in development and a documentary with a shoot in multiple locations in Italy. I also teach and dedicate a lot of time to updating that content to the latest workflows so research is inherent to my practice at every step. Half my day is research and development, prompt iterations, and workflows; the other half is unglamorous producer work: budgets, call sheets, emails, film festival submissions. Then I have client meetings and teach. It looks like magic in the final cut. It feels like project management with a render queue!
Where is your field in 12 months - - one specific prediction?
Within twelve months, AI production becomes a normal line item (broken down into a ton of sub-items) in branded and corporate budgets, sitting next to "editing" and "motion graphics" instead of hiding in an experimental R&D fund. I'd put money on mid-size brands and institutions requesting AI-native as the default for internal and social content, because the math on ten episodes versus one shoot is impossible to ignore. The traditional festival world will keep arguing about purity as some brave creative souls develop interesting hybrid films. The commercial world will quietly move on, the way it did with drones and phones-as-cameras. The clients who move first will pay less and ship ten times more.
Where should readers find you, and what's the first thing they should join or read?
Find me on LinkedIn and on my AI Cinema Substack, ranked #10 Rising in Film and TV. Start with the Substack: it's the clearest window into how this work actually gets done, method and all, with none of the hype. And if you want proof over theory, my film Alma Robot, a hybrid film shot in Patagonia under the mentorship of director Paolo Sorrentino, is coming out soon.
AI Central Voices is where the AI Central team sits down with the founders, executives, and builders shaping AI - going behind the scenes of how they operate, what they're betting on, and where the industry goes next.
Want to be featured, or have an event you'd like us to cover? Reach out at [email protected]
