Jason Day leads Luma’s EMEA business, where the AI research lab is building multimodal models and tools designed for creative work.
Luma Agents, the company’s creative intelligence platform, lets people create in any format while keeping hold of context and intelligence, whether they’re using Luma’s own models or third-party ones. For creators, Day says, that means moving at the speed of their imagination rather than getting stuck stitching different models and products together.
He sees that as especially important in EMEA, a region rich in languages and cultures and, typically, the place where most AI falls short. When linguistic, cultural or brand context isn’t retained, it gets flattened across different products and models. Luma’s approach focuses on keeping every form of context and intelligence intact, whichever model is being used within Luma Agents.
The company counts Publicis, Serviceplan, Adidas, Lufthansa and Arla among its partners, supporting agencies and brands to produce hyper-local, culturally relevant campaigns efficiently, without anyone needing to become a prompt engineer. In Day’s words, the shift Luma is driving is towards work that’s faster to make but never loses what makes it authentic to the people and places it’s meant for.
Day shared with us how Luma Agents is changing the way creative work is made, why video is at the centre of the company’s route to AGI and where AI-generated film is heading next.
We hear a lot of talk in the UK and European tech scenes about building an “intelligent world model.” What does an actual, functional world model mean to Luma, and why do you think so many companies trying to build them are getting it completely wrong?
At Luma, our mission is to build unified general intelligence that can generate, understand and operate in the physical world. A truly successful world model is one that fully understands how the world works.
To understand how the world works, we need to be able to interpret across every modality, especially video, which is Luma’s heritage. If we can truly assimilate what’s happening in the world, then we can apply the intelligence of AI to some of the world’s most complex problems.
How does Luma’s work in video generation lay a much simpler, more direct path to robotics and artificial general intelligence (AGI) than the rest of the industry is making it sound?
We believe the most direct path to AGI and advanced robotics is through video. Video inherently captures the physics, spatial dynamics and continuous movement of the real world, things that language can only abstractly describe.
By training models to deeply understand and generate video, we’re effectively teaching them how the physical world operates.
What has been the biggest challenge you’ve had to overcome along the way?
Launching Luma’s first commercial presence outside of the US has been an incredible experience. We’ve hired a great team, built a strong profile and have amazing client partners, but of course there are many challenges along the way!
One of the biggest hurdles is reframing how people think about AI. The clear evidence this year is that AI is an enabler of great talent, enabling them to deliver amazing craft and value to their clients.
Many initially feared that generative technology would erode creative jobs or destroy the craft, but the emerging consensus is that the technology is an enabler, not the enemy, of creativity. Great work happens when strategy, creative thinking and technology meet.
If we look three years ahead, what is the one capability or use case for AI video that currently doesn’t exist but will fundamentally change how films are made?
Three years is a very long time horizon. In the past six months, AI models have dramatically improved such that the reality of the technology is, perhaps for the first time, ahead of expectations.
In a lot of cases, AI-generated images and videos are already indistinguishable from “traditionally” shot formats. Within 6-12 months, that gap will continue to close to a point where it is impossible to differentiate. This throws up interesting questions around verification of content and brand safety that we should be collectively discussing across governments, industry and the creative community at large.
One thing I’m certain about is that AI is an enabler of incredible human craft, ingenuity and creativity. At Luma, we’re focused on supporting these people to be prolific, and that’s an exciting journey to be on over the next three years.
