What happens when everyone can stand on the rooftop and see the world in a completely different way?

At Vurvey Labs, that question wasn't just a thought experiment—it was the inspiration behind Project Rooftop, our groundbreaking collaboration with Unilever. We are thrilled to announce that this innovative initiative has won 1st Place at the I-COM Data Creativity Awards in the Artificial Intelligence category.

The award recognizes our unique approach to synthetic data: moving past static, robotic AI to create People Models that simulate true human behavior, emotion, and feedback at scale.

Here is the story behind the award-winning case study, and how we are redefining the future of consumer insights.

The Challenge: Speed, Scale, and the Flaw of the "AI Mean"

In a marketplace moving at breakneck speed, Unilever’s insight teams faced a massive business imperative: transition from episodic, one-off research projects into a continuous flow of real-time consumer intelligence. They needed to democratize access to these insights for global and local market teams alike.

However, as teams began experimenting with standard Large Language Models (LLMs), they hit a frustrating barrier: everything gravitated back to the mean.

Standard AI models are designed to find averages. But meaningful consumer innovation is rarely found in the average; it lives in the unique nuances, the contradictions, and the hyper-specific ways real people see the world. Unilever didn't need generalized data. They needed deep, diverse, human-centric intelligence at the speed of synthetic data.

The Breakthrough: Stepping into "Surreal Mode"

To solve this, Vurvey took a novel approach that bridges the gap between real-world human experience and synthetic scale. We call this middle state Surreal Mode.

What is Surreal Mode?

It is the synthesis of human and AI. Think of it like a dreamlike state where you play out ideas in your head to see how people might react. By merging qualitative human depth with AI-driven scale, we create "surreal worlds" where distinct consumer populations can come to life.

Through this approach, we pioneered a brand-new AI category: People Models.

Understanding People Models

While the tech industry has focused heavily on Content Models (like text and image generators) and World Models (spatial intelligence and robotics), people have been skipped over.

Vurvey’s People Models fill this gap. They are not just static personas; they are dynamic, unique agents that can watch concepts, talk back, and simulate real feedback. Instead of interacting with a single generic bot, we generate entire populations to ensure the rich diversity that brands need for authentic testing.

How It Works: Continuous Training, Not Static Data

Our foundation as a company is rooted in human emotion. Vurvey actually started as a shorthand for "video survey"—our initial patent focused on how to interview the world and capture raw emotion through life's micro-moments.

We brought that exact DNA into our AI. Our People Models are fed by a continuous stream of interviews with real consumers worldwide.

• Behavior Over Trends: We don't try to read the tea leaves on fleeting macro trends. Instead, we focus deeply on human identity and how identity drives behavior.

• Continuous Training: Traditional LLMs are often six months to a year behind due to rigid pre-training cycles. Vurvey's models undergo continuous training, introducing shifting consumer facets in real-time. As humans evolve, our People Models evolve with them.

• Privacy First: Every simulation is built with strict frameworks that keep customer data private, secure, and paramount.
  
  

  

  
  

The Results: Global Scaling and Provable ROI

The impact of this methodology across Unilever has been massive. By sitting side-by-side with Unilever’s teams to build out these validation frameworks, the internal trust in AI simulation skyrocketed.

• Massive Scale: Adopted globally across four of Unilever’s top product categories.

• Proven ROI: Deployed across more than 1,000 internal projects, resulting in clear, substantial cost savings and unprecedented speed-to-market.

• High Predictive Accuracy: In internal validation studies comparing real consumer interviews against our AI populations, we found strong concordance. The models successfully predicted real-world outcomes, such as whether a concept would sell out over a launch weekend.

  
  

AI Powered by People™

Winning the I-COM Data Creativity Award validates our core mission at Vurvey Labs: to build AI powered by People™.

Synthetic data shouldn't mean losing the human touch. By combining synthetic speed with the qualitative depth of real human identity, we are helping the world’s biggest brands innovate with unparalleled confidence.

Thank you to the I-COM judges, and a massive congratulations to the Unilever and Vurvey teams who stood on the rooftop to bring this vision to life!