INSIDE HOGARTH'S AI-POWERED CREATIVE PRODUCTION STRATEGY
Adapted from an Orange Logic webinar hosted by Misti Vogt. Questions have been consolidated and rewritten for clarity. Watch the full on-demand webinar below to hear directly from Chico Lima, Global Head of Emerging TV and Creative Technology at Hogarth, part of WPP.
Global brands are under pressure to produce more video content, in more formats, for more channels and audiences, without increasing production timelines or budgets. At Hogarth, a campaign that once required 200 versions can now demand more than 2,500. Each variation may require different aspect ratios, edits, creative elements, and regional adaptations, making traditional production workflows increasingly difficult to scale.
Chico Lima, Global Head of Emerging TV and Creative Technology at Hogarth, is helping rethink how creative teams use generative AI across the production lifecycle, from previsualization and creative testing to content adaptation and delivery. Rather than simply accelerating existing processes, Hogarth is exploring how AI can fundamentally change the way teams collaborate, make production decisions, and reuse content. With Orange Logic supporting asset management, metadata, and traceability, the goal is to bring greater speed and flexibility to production without losing control over creative quality, brand accuracy, or content provenance.
Chico Lima: The challenge today is the multitude of platforms we have to deliver content to. Several years ago, we would have an MP4 in 16:9, export it, send it to YouTube or Facebook, and that was it. Now we have all these different platforms, formats, aspect ratios, and specifications. For example, when we had a campaign for Nespresso, we would probably produce around 200 versions. Now we have more than 2,500 versions. Every time you change the aspect ratio from 16:9 to 4:3 to 1:1, you have to reframe, reformat, and sometimes change the edit itself.
Brands need to deliver the right message at the right time, on the right device, in the right context, in a way that adds value to the person receiving it rather than just bombarding them with information they don't need. They need to work at an unprecedented speed, while timelines are getting shorter and budgets are getting tighter. That makes everything much more challenging and complicated.
Chico Lima: When we're working on TV commercials, one of the most important things is being able to visualize what we're creating before we get on set. Traditionally, you start with a script, which everyone can interpret differently. Then you create storyboards, a fixed set of images that helps the client understand how the idea will come to life. In the early 1990s, we started seeing 3D previsualization, or previs. Suddenly, you had color, motion, and technical information about cameras, lenses, cranes, locations, and lighting. You could explore whether you needed a camera car, helicopter, or drone, whether to shoot in a studio or on location, and how the time of day would affect the shadows. All of that helped you make better production decisions.
Now, when you bring generative AI into that process, you can iterate and generate those visuals much more quickly. You can take an image of a car, generate a 3D model, bring a Google Maps 3D environment into the previs world, and start exploring different camera angles, lenses, and compositions. But AI takes that a step further. You can generate a photorealistic image from rough 3D, and from a single image, you can generate photorealistic video. In the time it previously took us to produce a storyboard, we can now have a photorealistic video. That changes the game completely.
The other important thing is that we can send these visuals to clients and give the entire team a much better understanding of the creative intent. We can also send different versions to focus groups, see which ones perform better, and give brands more information about where it's worth investing.
Chico Lima: I like to use an analogy from the Industrial Revolution. When factories used steam engines, all the machines had to be positioned around the engine. The entire factory was designed around that limitation. When electric motors arrived, many factories simply replaced the steam engine with an electric one. But it took about 20 years for people to realize that smaller electric motors could be distributed throughout the factory, allowing them to redesign the entire production process. With Henry Ford and the assembly line, they could produce and assemble cars 10 times faster.
I think about generative AI as that new engine. The question is which organizations will be able to leverage the power it brings to content creation. With transformative technologies like generative AI, the organizations that will benefit most are the ones willing to do the hard work of reimagining their systems entirely. It's not only about technology. It's about redesigning processes and principles, and most importantly, changing the mindset of the teams working together. Generative AI brings agility to the creative process because you can quickly iterate, prototype, experiment, and gather data. From the initial idea all the way through adaptations for social media, you can continuously generate content that's fresh and relevant to what's happening in the world.
Chico Lima: We envision cross-functional teams working together in the same room. You have a data analytics person looking at market trends, a creative director, someone from the brand, a director, a producer, people prompting AI, an editor, and a colorist. Everyone is collaborating and generating visual content almost instantly. Before, you would have to create previs, which could take several days, or rely on images from your archive or stock footage that didn't necessarily fit the creative intent. With generative AI, you can generate content on the fly. That changes everything. These teams can come together to ideate, prototype, send content to the client, get approval, gather information about what's performing, and then come back and iterate.
But the important thing is that the entire process has to change. If you can generate videos very quickly but then it takes too long to determine how to produce them, what they'll cost, or whether they'll fit the timeline, you've created another bottleneck. Every stage needs to be optimized and redesigned to work together.
Chico Lima: Although AI is extremely useful for pre-production and the latest models can generate some final shots at a high level of quality, there are many situations where a hybrid approach is better. For example, if we're creating an automotive commercial featuring an actor, we might shoot that actor on an LED volume, even if the background is generated by AI. We still want to capture the performance, particularly if we're working with a celebrity. For the interior shots, we might need the actual car in the studio. Everything has to be accurate and exactly on brand.
For exterior shots, perhaps the AI-generated footage is good enough to use. But if we're generating content at scale for personalization, we need much more control. The car might look real, but it might not be the exact model we're advertising. If we're producing different versions with different wheels, colors, and details for different regions or audiences, those details have to be correct. That's where the hybrid workflow becomes important. We might take an AI-generated shot and overlay the manufacturer's CAD data using CGI. That way, we can generate lots of different versions while making sure every detail, from the tires to the rims, is accurate. We cannot have AI hallucinating and creating features that don't exist on the car. That could cause serious problems.
For other shots, such as an object breaking into multiple pieces and scattering across a road, we might decide to use full CGI with procedural simulation tools like Houdini. It's about understanding the best approach for each shot. Once you can prototype and visualize what you want to achieve, you can put budgets together much more quickly and determine the right way to deliver the creative vision.
Chico Lima: When you think about your library of assets, generative AI opens up new possibilities for how you can use existing content. Let's say you have a shot that was originally created for a summer campaign. You can now use AI to generate a winter version. Suddenly, you start to rethink how you use your archive. You can go back and find shots that might not work for what you need today, but with AI, you can generate new versions that are relevant to your current campaign. That brings the power of generative AI into the creative process in a very practical way. You're no longer limited to using an asset exactly as it was originally produced.
Chico Lima: Modularity becomes more important than ever because you're able to combine all these different elements on the fly in a way that wasn't possible before. It's really important to think about that from the pre-production stages. That's one of the things we like about having everyone working together from the beginning. When you're in that room, you're not just thinking about the video. You're also thinking about the stills, all the additional content you need to capture, and the different elements you need to account for.
You have to think at the campaign level, from the main video and stills all the way through social media, and make sure you have all the elements aligned, available, formatted, and organized so you can deliver that content. That's especially important for social media, where you might need thousands of versions, combining elements in different ways depending on the audience you're targeting. The earlier everyone is involved in those decisions, the better positioned you'll be later in the process, and the more content the brand will be able to produce.
Chico Lima: We set up a program called AI in Motion, where we invite technology companies to present their latest AI innovations, solutions, platforms, or models to the wider organization. We organize global presentations across different time zones when possible, record the sessions, and evaluate the outcomes. We look at which technologies offer something we don't already have or do something better than our existing tools.
After that, we bring together a smaller group of people who are really passionate about AI and have hands-on experience through personal projects or client work. They test the technology, sometimes over three or four weeks. In parallel, we run the technology through WPP's AI governance team to validate security and safety. We also look at the terms of agreement, enterprise accounts, licensing costs, seats, and how the technology would connect with what we already have. That's our process for the time being. We're constantly updating, designing, and improving it as the technology evolves.
Chico Lima: Orange Logic plays a big part for us, and it's going to play an even more important role as we move toward these cross-functional production environments. One of the things that's really important is having a place where we can trace back how content was generated. Today, there are so many AI models and different technologies involved in creating content. You have lots of people working on a project, generating lots of different shots. Some of those shots end up in the final edit, while others are modified or replaced. You need to know who created each asset, which model they used, and even which prompt or seed number was involved.
That's the data we're working to bring into the system, so we don't just have the final assets and the associated After Effects or Premiere projects. We also have the traceability needed for audit purposes. There are challenges around AI copyright, but it's also about being able to quickly understand how something was made, who made it, and which platform they used. Today, someone might generate something online, download it to their Downloads folder, send it to someone else, and then it gets put into an edit. That makes everything very difficult to track.
That's where Orange Logic is integral for us. It allows us to organize content by project and maintain the metadata and traceability our producers need. This becomes especially important when you're combining AI-generated backgrounds with live action and CGI components. Without that organization, it can become a big challenge. We want to make sure everything comes together and that we can trace every asset back to how it was created.
Chico Lima: Yes. We're looking at building agents with specializations for different areas of production. When you're generating different versions very quickly to test with focus groups and clients, once the client likes one of them, you need to understand very quickly how you're actually going to deliver it. Today, that can involve calling lots of different people just to get an idea of the production approach.
We're building agents that can look at a video or a script and help determine the right approach for each shot. They can recommend options quickly, give you an approximate understanding of what each approach would cost, and help you see whether it fits the budget and timeline. That's an important part of pre-production. If we've redesigned the creative process so we can generate videos very quickly, but it still takes too long to determine how to produce them, what they'll cost, and whether they're feasible, then the process breaks down. That's why everyone has to be involved, and every stage needs to be optimized together.
Chico Lima: There are so many areas where you can make improvements that might not seem that exciting, but they'll streamline the whole process and allow you to get to the most valuable work faster. For us, one of the most important things is making sure the people actually doing the work get hands-on experience with the technology. That's one of the reasons we created AI in Motion. There were so many conversations happening at the leadership level with technology companies, but the people who would actually have to use AI to generate and deliver content weren't necessarily getting that experience.
So we started bringing technology founders in to demonstrate their tools, putting those tools in people's hands, and encouraging them to test. We engage directly with technology vendors and create safe sandboxes where people can experiment. At the same time, we're looking at the workflow or pipeline, understanding where things are getting stopped or blocked, and seeing where automation or AI can help. There's a lot of basic automation that can improve the process before you even get into more advanced AI.
But the most important thing is giving the actual creative talent access to the tools so they can play, experiment, and provide real feedback about what would make their lives easier. Whenever I talk about emerging technologies, my focus is on enabling the creativity of the teams doing the work. The more creative they can be, the better the outcome. Give people a way to safely test these tools and go from there. Don't overthink it. Of course, you still need to address security and compliance, and we do that in parallel. But people need to understand what's possible. Otherwise, they won't know which questions to ask or how to think about using the technology.