How VaynerMedia Is “Applying Principles of the Creator Economy to AI”

Consumers today expect brands to converse with them, not speak at them. They want content that reflects their culture, their humour, their interests, and the micro-moments that shape their daily lives. But delivering that level of hyper-personalised content at scale and across multiple markets is traditionally slow and expensive.

That’s the challenge Grab (Southeast Asia’s leading taxi and food delivery app) and VaynerMedia Asia Pacific set out to solve. The result? CreAItors– an AI-powered creative platform built to transform a single brief into thousands of culturally resonant stories. 

Its first activation is a campaign built to drive awareness and adoption of Grab’s Group Order feature. This was delivered through over 16,000 unique permutations across six Southeast Asian markets, proving that brands can now deliver relevance at the speed and volume today’s audiences demand – without costing huge amounts of time and money.

Powered by five key creative levers (location, occasion, cast, AI persona and platform format), CreAItors works by blending generative AI with data, platform insights, and human creative direction to produce hyper-specific characters, moments, and narratives to fit any demographic targeted.

To find out more, LBB’s Sunna Coleman spoke with VaynerMedia’s APAC MD, Tim Lindley who dives into how the technology was built, getting the balance between AI and human creative, and how they plan to expand CreAItors’ capabilities.

LBB> What was the core strategic problem that led VaynerMedia and Grab to develop CreAItors?

Tim> The core problem was a mismatch between the complexity of our audience and the limitations of traditional advertising. When a group of friends or coworkers order food on Grab, the combinations are mathematically infinite. Even within a single small group, behaviours and tastes are hugely diverse.

We famously champion finding relevance with specific cohorts rather than broadcasting to generic audiences. With Grab’s Group Order feature, we needed to speak to the ‘Keto King’, the ‘Lactose Dodger’, and the ‘Level 10 Spice Champion’ individually. The challenge was simple but massive: how do you achieve that level of relevance at scale within a realistic budget?

We developed CreAItors to address that volume and variety. They mirror the diversity of Grab users with a diverse cast of gen AI characters, allowing us to create social assets that feel like inside jokes tailored to specific personality types.

LBB> What AI technologies or frameworks power CreAItors – are they proprietary or built on existing models?

Tim> We built a workflow called the ‘Content Kitchen’ – essentially a gen AI version of our codified agency operating model. The tech stack at the time leveraged a number of best-in-class models, including Gemini for trend analysis, OpenAI for character development and scripting, and Kling.AI for video generation.

It’s worth noting that while those models evaluated well at the time, AI is a moving target, and new tools were dropping even as we were in production. So what felt cutting-edge in May already looks different today… and that’s the whole point. We are platform-agnostic, and the stacks will continue to change as technology evolves.

The real breakthrough here wasn't just generating text or video; it was the democratisation of high-end production. In the past, creating animated 3D characters required complex CGI workflows that were expensive and slow. By leveraging this gen AI stack, we bypassed those barriers.

We produced high-fidelity 3D characters and distinct worlds without the prohibitive costs or timelines associated with traditional animation. This allowed us to move much faster, making adjustments in minutes that may have previously taken days.

LBB> How do you see this approach reshaping the traditional creative agency model?

Tim> The traditional model was built to push one message to the masses based on a "common denominator" theory, but the platforms that gate-keep attention today – TikTok, Meta, YouTube – flipped that around. They leveraged technology to build highly personalised feeds, proving that we all like different things, and gave us spaces to consume only what is relevant to us.

The modern agency has to operate differently. We need to design for relevance, at scale, and at speed. In practice, this means understanding how platforms work, making more, and making faster.

A modern agency has highly leveraged ‘human + AI’ teams that connect the dots between ideas and distribution. AI isn’t a silver bullet, but it is an incredible enabler for talent, especially as the lines between creative, strategy, production, and media continue to blur – and we believe it's imperative to bring them all together. Creatives must become obsessed with performance and make content every day. Strategists need to channel relevance and turn complex data into immediate action. Technologists need to fine-tune models and build automated workflows.

AI is making this happen faster, and at a scale that traditional models cannot reach.

LBB> Can you explain how CreAItors blends generative AI with human creative direction in practice?

Tim> We applied the principles of the creator economy to AI.

We didn't just ask AI to make us some characters; our creative team carefully crafted the inputs, including tone, taste, personality, and more for each CreAItor. We treated them like social media influencers, training them to understand the audience they're speaking to, and to operate with a consistent "beat”.

In practice, the human team provided the soul, and the AI provided the body.

Our team maintained human control over the message and the strategy, while using AI to handle the heavy lifting of execution. It allowed our creatives to play a high-level editorial and directional role, rather than getting bogged down in the rendering process.

LBB> How do you ensure cultural nuance and accuracy in AI-generated content across different Southeast Asian markets?

Tim> We retain nuance by treating AI as a tool, not a decision-maker. The CreAItors were trained to resonate with very specific cohorts, and those cohorts can look very different across markets.

We established a “human firewall" around the AI. Before a single prompt was written, local teams defined the specific social archetypes relevant to their markets. For example, a "polite decliner" cohort in one culture will differ vastly from a "loud negotiator" cohort in another.

Language is a complex variable across this region, not just at a high level between Bahasa and Thai, but deeper into tonality, formality, and slang. We implemented back-translation steps using different models, allowing the human team to assess whether the desired tonality was retained throughout the workflow.

Because these characters have a specific personality, we could train the output to align with local slang and humour. But ultimately, accuracy comes from local oversight. Our in-market creatives manage the AI output to ensure the characters stay true and relevant to the desired cohort rather than falling into generic tropes.

LBB> What metrics or early results have indicated success (e.g. engagement lift, awareness, conversion)?

Tim> Grab has robust data that shows substantial uplift in media efficiency, CTR and conversion rates. Most importantly, they saw a huge increase in Group Orders attributable to the campaign.

LBB> You mentioned this is just the beginning – what’s next for CreAItors?

Tim> CreAItors isn’t a "campaign" with a start and end date; it’s a capability we have built for the brand.

Because we have codified the workflows and character profiles, we have effectively built the beginning of an owned IP library and a prototype executional model for Grab. We can reactivate these characters instantly for future briefs, deepen their lore, or introduce new avatars to match new consumer behaviours. The engine is built for scale and flexibility.

Secondly, we are constantly upgrading the tech stack. The tools we used to launch this are already being outperformed by newer models, which is the most exciting part of this space. The quality of output we achieved here is the floor, not the ceiling. Future iterations will benefit from higher fidelity and sharper storytelling as models evolve.

This project served as a proof of concept for a modern marketing engine. It has opened the door to a much wider conversation with Grab as to how we operationalise AI across multiple workstreams to drive efficiency and relevance across their entire business.