top of page

Maya: The New AI-Driven Frontier of Influencer Marketing

Artificial intelligence has completely redefined the world we live in today. Now, our mornings start with personal assistants waking us up and laying our favorite playlists. Bots help customers with their questions and issues about the services and products they have purchased.

And now, artificial intelligence is capable of redefining influencer marketing with the power

of creating a persona. Meet Maya — the next-generation virtual influencer set to transform the concept of brand marketing.

Everything started in 2019, when Puma faced the challenge to create a marketing campaign that would promote their “Future Rider” shoe to the audience in South East Asia. The core of the challenge was the complexity of the region in terms of their unique perspectives, cultures and influencer markets, making it virtually impossible to choose a single person to represent the diversity of the overall region.

Here is where and Ensemble Worldwide (part of IPG Media Brands) stepped in: turning to the power of artificial intelligence, the partners began strategizing on ways to create the persona from scratch by combining the unique characteristics of each region into one character in order to face the challenge of fragmentation.

In collaboration with SoMin, a social media artificial intelligence platform, the agency researched and mapped millions of Southeast Asian faces by analyzing social media users in APAC and applying neural networks to create Maya’s appearance.

And the result? Maya.

But it’s important to take a step back and take a look at the technology that enabled the creation of a virtual influencer telling people the stories of brands and better yet — showing and personifying them.

SomLin applied generative adversarial networks, or GANs, to analyze data and create Maya’s facial composition. Generative modeling is a type of unsupervised learning in machine learning that allows algorithms to automatically identify and learn patterns in the data they are given so that the model can generate new examples of outputs that could have been inferred from the original dataset.

GANs are a quickly evolving and growing subfield of machine learning due to their capacity to generate realistic outputs, especially when it comes to image processing and translation, as well as creating photos of objects, backgrounds and even people that can be hard to label as fake.

With the capabilities of the neural network in mind, SoMin trained the GAN to analyze the images and profiles of South-East Asian lifestyle and fashion influencers that attract consumer attention across regions. The model then started combining these profiles to generate an appearance that was a perfect blend of all the unique regional idiosyncrasies found across Southeast Asia. SoMin’s team created a few facial profiles that were presented to the client to select one fit for further production. Once the client settled on the final look, Maya’s personality was further fine-tuned to express emotions and versions through AI.

Now, Maya could look happy or sad, older or younger. Just like a human.

Creating a virtual personality, however, didn’t end with looks: it was also important to create a relatable personality with a backstory. Maya’s personality was crafted based on proprietary UM Studio wave studies on Southeast Asian psychographics; it was designed to evolve with time through social listening, as the AI algorithms studied Maya’s interactions online and curated patterns for her to follow. The ultimate goal was to create a persona that would portray Southeast Asian values and be as relatable to the audience as possible.

The final puzzle piece is who Maya was supposed to be is the fact that she doesn’t look completely realistic: instead, there is a slight aspect of a video game character to her final appearance, almost as if it makes the viewer pause for a second in disbelief, creating an air of suspense around her.

Maya is only one example of how AI will completely transform brand initiatives and consumer-brand relationships in the future. It takes brand storytelling to a new level, allowing new alleys for consumers to experience the stories rather than read about them or listen to them. The technology opens a new door for companies to experiment with the touchpoints that consumers can interact with, fostering and creating new and deep relationships between brands and their loyal followers.

Maya is also a sign of a world that is heavily digital; especially after the pandemic, the shift towards the digital space across all activities — both on an individual and corporate level — and how the virtual space has provided a new reality for people and brands to communicate and interact.

The existence of a virtual influencer may be a new beginning, and with opportunities, there also come questions: how “conscious” will virtual personas become? How will they find their unique position in society? And what does this mean for real-life, human influencers?

Only time will tell.

Originally published in Medium


bottom of page