Bloomberg Tech

Caroline Ingeborn on hybrid filmmaking, not prompt-to-movie

Caroline Ingeborn· COO at Luma AI
·~7 min·English·Bloomberg
MultimodalAgentsAI CompanyBusiness Strategy
TL;DR

Luma AI's COO Caroline Ingeborn argues the future of AI in film isn't a prompt that spits out a movie but a hybrid craft: route each shot to the best model at the right token cost, keep the actor's performance, and bring real production back to Los Angeles.

01Company Structure

A three-layer company, not a model shop

Luma isn't just a model shop: it's a three-layer stack of multimodal models at the base, products that let creatives use them with control, and forward-deployed people who make adoption actually happen.

And the third layer is having forward deployed creatives and engineers so that adoption really actually takes place.

— Caroline Ingeborn, Bloomberg Tech
Key Insight
The third layer is the tell: Luma treats adoption as a people problem, not only a product one. The implication is that a capable model is the starting point, and deployed creatives and engineers are what turn it into real usage.
02Core Mental Model

Not prompt-to-movie, a tool

The prompt-to-feature-film dream isn't the goal: AI is a tool inside a collaborative craft, the latest in a long line of technology shifts Hollywood has always absorbed.

this is a tool as all other tools that we have seen, and Hollywood is, like, built on technology shifts.

— Caroline Ingeborn, Bloomberg Tech
Key Insight
Calling AI a tool is positioning as much as description: it places Luma alongside filmmakers rather than against them. The implication is that adoption turns less on raw model power than on fitting into how creative work already happens.
03Model Routing

No model to rule them all

There's no single best model, so Luma's creative agent routes each task to the right one: the top video model when quality matters most, a lighter and cheaper model for high-volume ideation.

there is no model to rule them all. There is no model that can make a movie even though I don't think any video AI company has that as a goal.

— Caroline Ingeborn, Bloomberg Tech
Key Insight
Routing by task turns model choice into cost control: reserve the premium model for the work that needs top quality and use cheaper models to explore. It also makes Luma model-agnostic, a hedge against any single lab's lead.
04Industry and Economics

Budget is the real constraint on a story

Ingeborn frames budget, not ambition, as what limits storytelling: a director shot years of TV abroad to stretch his investment, and hybrid production is Luma's answer to that constraint.

So he spent years making TV shows abroad because that's how he could get the most out of the investments that he was doing and to tell his story.

— Caroline Ingeborn, Bloomberg Tech
Key Insight
Naming budget as the constraint reframes AI's role as a cost lever, not a replacement. If cheaper production is what lets more stories get made, the argument that work returns to LA follows; whether it actually does is the open question.
05The Method

Is it AI? No, it's hybrid

Hybrid filmmaking keeps real actors on a real stage and uses AI to place them in any world, retaining and even enhancing the performance rather than replacing it.

And this, like, retains the actor's performance. It enhances it.

— Caroline Ingeborn, Bloomberg Tech
Key Insight
Because hybrid keeps real actors on set, it frames AI as augmenting a performance rather than producing one from nothing. The implication is that the people on a production stay the creative anchor while AI changes the world around them.