Multi-Modal GenAI Is Transforming the Creative Industries, Discover How From a Visionary CEO — And What Comes Next

In June 2024, Toys”R”Us launched the first fully AI-generated brand ad. Other brands, like Motorola, soon followed suit and now multi-modal generative AI (GenAI), which works across content types — e.g. turning text into video — is driving seismic change in the ad industry with its speed of creative production. 

It’s not surprising then that Boston Consulting Group research shows 80% of marketing, sales, and service leaders expect to boost GenAI investments in the next 12 months. One challenge with that is, as PYMNTS Intelligence shows, 42% of CMOs struggle to understand its full potential within their organisations.

To avoid missing the opportunity, it’s important to know how industry leaders are making the most of this multi-modal moment. To find out, we spoke to Oliver Yonchev, Co-founder and former Chief Executive Officer of marketing agency, Flight Story. 

The democratisation of learning and making 

While the focus of the discussion on GenAI is often on what we make with it, Oliver believes how we work with it isn’t talked about enough. “We all learn very differently and we all have different working styles. Some people are optimally effective when they listen to information, some people prefer to read it.” 

He uses himself as an example, “If I could have a briefing in audio format while I was on my way to a meeting, that would be more useful to me than sitting and reading a 40-page report.” 

He adds that GenAI isn’t just useful for helping people receive information; it helps them express their ideas too, stating, “It ends the friction between thought and digital execution.” 

What does this mean for marketers specifically? “We now have new ways of presenting information to clients.” He describes creating a podcast for a pitch, “previously that’d be impossible, or a very costly exercise. Now you can do it within five minutes on an application.”  

Multi-modal GenAI is a problem-solver 

The solutions that multi-modal GenAI provides are sometimes unexpected. Oliver tells the story of when a project stalled due to a speaker pulling out of a film. “We were under quite a tight timeframe, so we didn’t have a choice to move the production back. So we went remote and used a virtual speaker, but on the day that footage was corrupted.”  

A serious setback, but they got creative with GenAI. “We effectively created an avatar of the speaker, used voice synthesis to emulate the voice, and we embedded the sections within the production.” Great news for clients’ budgets. “Without GenAI that would have meant a full-day production. It would have delayed all the launch timeframes.”  

This illustrates the pace of progress, “We were capable of doing something that we wouldn’t have just a year ago,” he says. 

Multi-modal GenAI is a content multiplier 

While half of CMOs use GenAI for automating tasks, according to PYMNTS Intelligence, Oliver thinks the value goes far beyond that, saying, “I think of generative AI as a content multiplier … if you’re a global e-commerce business, you have an ungodly demand for content and resizing and images. Now you can do that extremely cost-efficiently.” 

As well as scaling content creation, it scales content comprehension. “Before if you wanted to analyse 100 videos, it would be a very time-consuming task,” but now Oliver says it can be performed by a machine with the information instantly interrogated: “Tell me five themes that all these top 100 performing videos share, tell me the most common hooks used, tell me what’s the most used colour, you can keep going.” The advantage being, “real-time insight and intelligence that can form your decision making.” 

Multi-modal now; agentic processing next 

Oliver believes we’re in the middle of a three-phase evolution of GenAI. “Phase one of the generative era was chatbots; phase two was multi-modality; the next phase then is agentic processing.”  

Agentic processing is where AI systems can autonomously complete tasks by interacting with other AI systems, and, according to Oliver, “It’s a term you’re probably going to hear more and more in maybe the next 18 months.” With this he sees the role of AI “moving from being kind of this passive collaborator with people to actually being an active collaborator in an organisation.” 

Oliver describes how agentic processing could be used to write a blog. “We could create an agentic process which starts with an AI transcribing a call. We use another AI that’s been trained on a specific tone of voice to then write an article. Next, another AI optimises that for SEO by going online and searching for keyword trends and then embedding them. Finally, another AI publishes it.” 

Challenges ahead 

Ethical challenges of AI have been much discussed, but for Oliver, something holding back that conversation is the lack of a common definition of what AI generated means. But don’t expect one soon. “No one knows what they need to do. Legislators don’t know what they need to do, so it’s impossible for a company.” That means that organisations must, for now at least, look within for the answer and “draw some sort of moral guidance around their own boundaries.” 

While a tech-optimist, Oliver has considered the unintended consequences of GenAI for our abilities. He suggests, “You could argue one of the things that makes you a really good writer or creative or a great marketer is doing a lot of the boring administrative stuff over and over again. If you remove those aspects, are people still developing the necessary skills to be a really good marketer?” 

He’s philosophical on this. “The technology’s neither good nor bad. It’s there. Many amazing things will happen, but there’ll be some negative things as well.” 

Making the most of multimodal GenAI ultimately comes down to education  

While some organisations are experimenting with all of the above, many aren’t getting the results they expected. “You have large enterprise companies that have rolled out ChatGPT licences across their teams, at great expense, and find the utilisation is 20%.” 

For Oliver, the problem is not the users but the training. “They’re not being taught to prompt engineer … we’ve handed everyone these really powerful tools and just assume everyone can use them.” Which they won’t be able to, because our old skills are redundant now.

“Chat interfaces are very different. We’ve all learned some bad behaviours from search. We’ve been taught to use keywords; now with the chatbot interface, we have to be really specific.” And it’s not just a new set of instructions. “You need to know how to critically think — it’s a different skill in getting good outcomes.” 

Already, the chances multi-modal GenAI provides for marketers are many, and many more will be discovered yet. Ultimately, for Oliver, “It’s not really about changing what we do as marketers, it’s changing how we do things.”