Executive Overview
The landscape of human-computer interaction is undergoing a profound paradigm shift. For decades, digital devices have been fundamentally blind to the internal states of their users. They register clicks, track keystrokes, and measure swipes, but they remain utterly oblivious to whether the human behind the screen is deeply focused, cognitively fatigued, profoundly stressed, or even within a legally protected age bracket. Enter HarmonEyes, a pioneering artificial intelligence company co-founded in 2024 by serial technology entrepreneur Adam Gross and renowned researcher Dr. Melissa Hunfalvay.
At the heart of the company’s technological ambitions is Theia, an advanced AI foundation model designed to infer aspects of a person’s cognitive, emotional, and physical state entirely from real-time eye movements. By leveraging edge computing and standard device cameras, Theia promises to unlock a new era of "human-state intelligence." Its potential applications span a vast array of industries, including healthcare, automotive safety, extended reality (XR), defense, education, industrial safety, gaming, and market research.
Among its most urgent and heavily scrutinized commercial applications is age assurance. As regulatory bodies worldwide—from the UK’s Ofcom to the European Union’s stringent data protection authorities—crack down on digital safety, platforms are scrambling for reliable ways to verify user ages without compromising user privacy. Unlike traditional behavioral profiling, which builds intrusive dossiers by tracking a user’s habits, browsing history, and content preferences over time, HarmonEyes’ physiological approach seeks to verify age in seconds through passive, non-invasive eye-tracking features.
Operating strictly at the edge and discarding raw data the moment an output is generated, HarmonEyes is positioning Theia not merely as another surveillance tool, but as a privacy-first infrastructure play capable of reshaping how humans interact with technology across operating systems, app stores, and consumer applications.
Detailed Chronology: From Clinical Eye-Tracking to Edge-Based AI Intelligence
The origins of HarmonEyes trace back well before its official incorporation in 2024. Its intellectual and technical roots lie in RightEye, an eye-tracking technology platform co-founded by Adam Gross in 2013 alongside Dr. Melissa Hunfalvay. Over more than a decade, RightEye developed sophisticated eye-tracking solutions deployed across healthcare diagnostics, professional sports performance optimization, and military tactical assessments.
Gross’s entrepreneurial journey spans nearly three decades of scaling technology-driven businesses across healthcare, financial services, and data analytics. His track record includes co-founding KeyClaims, a healthcare cost-containment company acquired by Stratose/Zelis Healthcare in 2012, and financial analytics platform Trending123.com, acquired by InvestorPlace Media in 2007. He also founded financial technology firm KeepMore and co-founded Eastern ATM, an early East Coast private ATM network acquired in 1998.
When Gross and Dr. Hunfalvay launched HarmonEyes in 2024, they sought to distill this accumulated expertise—spanning millions of records and billions of data points—into a scalable, generalized foundation model. Three major industry shifts convinced the founders that the timing was right to bring human-state intelligence to billions of devices:
- Ubiquitous Camera Hardware: Ordinary consumer devices—smartphones, laptops, and tablets—now come equipped with high-resolution front-facing cameras capable of capturing subtle gaze vectors.
- Advances in Edge Computing: Modern processing chips can run complex machine learning models directly on the user’s device, mitigating the latency and privacy vulnerabilities of cloud-based processing.
- Regulatory and Societal Pressure for Privacy: Increasing legislative demands for age verification and user safety have collided with a growing public backlash against invasive data harvesting and behavioral profiling.
By combining RightEye’s proprietary archive with modern deep learning techniques, HarmonEyes built Theia, bridging the gap between specialized medical-grade eye tracking and mass-market consumer software.
Supporting Context & Metrics: Decoding Theia’s Technical Foundations
To understand the scale of HarmonEyes’ technological achievement, one must examine the underlying data architecture. The company’s models draw upon a massive proprietary dataset derived from RightEye, encompassing more than 585 billion eye-movement data points across 16.3 million records. This vast reservoir of physiological data serves as the foundation for training and refining Theia’s machine learning algorithms.
How Theia Works: A Technical Walkthrough
When a developer integrates Theia’s software development kit (SDK) into an application, the processing pipeline unfolds in rapid, discrete stages:
- Gaze Vector Capture: An ordinary device camera captures the user’s face, calculating gaze vectors to determine precisely where the user is looking.
- Feature Extraction: The SDK processes these gaze vectors, converting them into fundamental eye-tracking features such as velocity (how fast the eyes are moving) and fixations (where and how long the eyes are focusing).
- Machine Learning Inference: These generated features are fed into HarmonEyes’ proprietary ML and AI algorithms to infer cognitive load, attention, emotional readiness, or age ranges.
- Immediate Data Erasure: All of this processing is edge-based—occurring directly on the user’s device—and raw eye-tracking data is discarded immediately after each output is generated.
Empirical Validation and Accuracy
In published research evaluating the model’s age-estimation capabilities, HarmonEyes analyzed eye movements from 45,696 individuals, reporting a 94.67% classification accuracy across 12 distinct age groups.
Gross clarifies what this accuracy figure actually measures: "The accuracy measures any person’s data added to the dataset, and 94.67% of the time that person’s eye movements will correctly categorize the age group they are in (out of 12 groups)."
However, the company remains transparent about current limitations. The initial study deliberately excluded individuals with specific vision disorders or neurological conditions. Consequently, ongoing validation efforts are expanding to include populations with non-functional eye movements to ensure the model performs robustly across diverse demographics.
Official Statements: Adam Gross on Privacy, Ethics, and the Future of Age Assurance
In an in-depth interview, Adam Gross addressed the critical distinctions between physiological age estimation and traditional behavioral profiling, as well as HarmonEyes’ strategy for regulatory compliance and defense against circumvention.
Behavioral Inference vs. Physiological Age Estimation
Regulators like the UK’s Ofcom have drawn sharp distinctions between behavioral age inference—which observes how a user interacts with a platform over time—and explicit age checks. Gross explains how HarmonEyes navigates this divide:
"Our technology only requires a small sample of eye movements from a subject to place an individual in an age range or to confirm someone is younger/older than a specific age. We are able to collect eye movement data passively, continuously, and non-invasively without storing or collecting any personally identifiable information. Furthermore, all eye-tracking data is destroyed once the outputs are delivered."
Ethically, this approach breaks away from standard surveillance capitalism. "We are not profiling a person’s behavior, habits, decisions, or preferences—unlike behavioral inference, which by definition builds a picture of someone from how they act over time," Gross emphasizes. "We deliver age assurance without collecting or retaining that kind of personal information, so users and parents can be confident they are being protected rather than tracked or profiled."
Defending Against Spoofing and Circumvention
As digital verification tools proliferate, bad actors inevitably seek ways to bypass them. Gross notes that all HarmonEyes age assurance models incorporate built-in liveness detection and spoofing checks:
"Eye movements are unique to an individual, and HarmonEyes is able to determine if the eye movement for an account matches the person who set up the account, or is a different person—an older sibling, for example."
By analyzing the unique micro-movements associated with live human gaze, the system can thwart attempts involving prerecorded videos, photographs, synthetic faces, or unauthorized users attempting to take over a check.
Handling Uncertainty and Layered Security
No verification model is infallible, particularly when users sit right on the legal threshold (e.g., turning 16 or 18). Gross outlines how developers should handle probabilistic outputs:
"No age estimation model is perfect. We provide a confidence level with every result. This level of accuracy and confidence may be enough to confirm age assurance. However, if for example the confidence level for a specific measurement is below a certain level, it could be used as a way to trigger a secondary verification."
In high-stakes environments, Theia can function as part of a layered security architecture, acting as a rapid initial risk signal that escalates ambiguous cases to alternative verification methods only when necessary.
Architectural Verification and Regulatory Compliance
To satisfy stringent global frameworks—such as the European Union’s push for anonymous proof-of-age systems and the UK’s rigorous digital safety standards—HarmonEyes is committed to absolute architectural transparency.
"We recognize that architectural claims are only as credible as the ability to confirm them externally. We are pursuing third-party certifications and attestations designed to let developers, auditors, and regulators verify that eye-tracking data is not collected or retained in deployed applications," Gross states.
Furthermore, every model shipped by HarmonEyes is accompanied by a comprehensive white paper detailing its development methodology, accuracy metrics, and known limitations. The company also makes Theia available for live, remote testing so regulators can evaluate real-world performance firsthand.
Future Outlook: Where Human-State Intelligence Goes Next
As age assurance and digital safety regulations migrate from individual platforms toward operating systems, app stores, and core device infrastructure, HarmonEyes is positioning Theia to integrate seamlessly across multiple layers of the technology stack.
Gross envisions Theia operating within:
- Private Tech Stacks: Embedded directly into social media platforms and digital communities.
- Device Infrastructure: Integrated at the operating system level across smartphones, tablets, and laptops.
- Commerce Infrastructure: Embedded within payment networks to verify age-restricted transactions (e.g., gaming, alcohol, or financial services).
- App-Level SDKs: Deployed natively by individual consumer application developers.
Responsible adoption, according to HarmonEyes, hinges on three core pillars: uncompromising data privacy, transparent validation processes, and cost-effective edge deployment. By proving that advanced biometric intelligence can operate without centralized data harvesting or invasive surveillance, HarmonEyes is establishing a new benchmark for ethical AI.
As digital spaces face mounting pressure to protect minors and understand user context without sacrificing civil liberties, technologies like Theia offer a glimpse into a future where our devices understand how we feel and what we need—all while preserving our fundamental right to privacy.
To learn more about the future of human-state intelligence and explore integration opportunities, visit HarmonEyes.
