In a world where a viral TikTok video can send a brand trending globally in a few hours, the traditional market research cycle is becoming a liability. For decades, Fortune 500 decision-makers have relied on research cycles that often span 12 weeks. By the time a survey question is drafted, respondents are polled, and the data is formatted into a slide deck, the geopolitical or economic shift that triggered the study has often already evolved. The data isn’t just old; it’s frequently outdated.
Enter Brox, a predictive human intelligence startup that’s attempting to collapse those 12-week cycles into a matter of hours. By creating a “parallel universe” populated by 60,000 digital twins of real people, Brox allows enterprises to run unlimited experiments and simulations without the risk, cost, or time associated with traditional human panels.
Digital Twins vs. Synthetic Personas: Avoiding “AI Slop”
To understand Brox’s approach, it’s important to distinguish between a synthetic persona and a digital twin. Many AI companies create “digital audiences” using Large Language Models (LLMs) to generate generic personas. However, Brox CEO Hamish Brocklebank argues that this method produces “AI slop.”
Purely synthetic audiences tend to cluster around a narrow distribution of answers. They often over-index for “correct” or “healthy” behaviors—like eating broccoli—due to the fact that of the inherent biases in the underlying AI models. They simulate how an AI thinks a person should act, rather than how a real person actually behaves.
Brox takes a different path. Their digital twins are one-to-one behavioral replicas of actual, living individuals. The company recruits real people, pays them for exhaustive interviews, and captures deep demographic profiles and consumer preferences through a fully consent-driven framework.
How Brox Builds Behavioral Replicas
The fidelity of Brox’s technology comes from the sheer density of its input data. Rather than relying on a few survey answers, the company employs a rigorous data collection process:
- Deep Interviews: Each participant undergoes hours of real and AI-driven interviews.
- Psychological Mapping: The process digs into “decision drivers,” including upbringing, relationships, and marital stability.
- Massive Data Sets: For some twins, Brox maintains up to 300 pages of text data, creating what Brocklebank describes as the deepest per-person data set in existence.
To solve the “black box” problem common in AI, Brox uses a reasoning chain. When a digital twin predicts a specific reaction—such as how a person with a $2 billion net worth would respond to an interest rate hike—the model provides a step-by-step explanation of its logic. This gives clients the “why” behind the prediction, not just the “what.”
High-Stakes Applications in Finance and Pharma
Currently operating in the US, UK, Japan, and Turkey, Brox has digitized high-value cohorts that are notoriously tricky to access, including dermatologists and individuals with a net worth exceeding $5 million.
The platform is primarily being used in two critical sectors:
Pharmaceuticals
Companies utilize the platform to predict vaccine hesitancy or simulate how physicians might react to new biologics in response to shifting political climates. For example, the system can simulate how a public statement from a political figure might influence vaccine uptake.
Finance
Major banks use the twins to simulate depositor behavior. They can query the digital population to see if depositors at a major institution would move funds in response to geopolitical events, such as conflicts in the Middle East, or specific personnel news.
The Economics of Predictive Intelligence
Brox operates as a high-end Software-as-a-Service (SaaS) platform. Moving away from the per-project pricing of traditional firms, they offer enterprise-level commercial licensing with blanket flat fees. This encourages a culture of “testing everything” since clients aren’t penalized for the number of simulations they run.
- Entry Level: Subscriptions start at a minimum of $100,000 per year.
- Top-Tier: Global deployments for multiple teams can scale up to $1.5 million per year.
- Usage: Unlimited usage during the contract period.
Interestingly, Brox has also innovated how it maintains its data. Because high-net-worth individuals aren’t typically motivated by small cash payments to provide updates, Brox has issued Stock Appreciation Rights (SARs). This turns the participants into stakeholders in the company’s success, ensuring the digital twins remain updated and high-fidelity.
Why Prediction Markets Fall Short
The rise of prediction markets like Polymarket and Kalshi has drawn attention to the ability of crowdsourced betting to predict election outcomes or global events. However, Brox leadership maintains a distinct distance from this model.
Brocklebank argues that even as betting markets can predict if something will happen, they offer zero utility for business strategy because they lack the psychological nuance. Knowing there is a 60% chance of an event occurring doesn’t help a company adjust its consumer strategy; knowing why a specific cohort feels anxious does.
Key Takeaways
- Speed: Brox reduces research cycles from 12 weeks to a few hours.
- Fidelity: Uses 1:1 behavioral replicas based on deep interviews rather than synthetic LLM personas.
- Explainability: Employs a “reasoning chain” to show the psychology behind every prediction.
- Scale: Currently maintains 60,000 twins with plans to expand into the Middle East and APAC.
Backed by investors including Scribble Ventures, Wonder Ventures, and Vela Partners, Brox is betting that deep, proprietary human data will create a more resilient moat than commoditized AI models. As they expand, their ultimate goal is to simulate the world as a parallel universe, allowing for entirely risk-free corporate decision-making.