Trusted globally as a pioneer in privacy-first AIoT solutions
Inside
Emotion OS
Multimodal IntelligenceProprietary Multimodal Fusion. Fuses voice patterns, ambient sound, and physiological data directly on-device. This privacy-first approach captures robust emotional context even in noisy environments.
Digital BiomarkersData-Driven Emotion Modeling. Transforms raw physiological signals into structured, interpretable digital biomarkers using in-house algorithms, enabling deep understanding without subjective labels.
Long-Term MemoryRAG-Powered Evolution. Built on Retrieval-Augmented Generation to move beyond transient sensing. The system learns across time, allowing insights to become more personalized and accurate.
Deployment & ArchitectureScalable On-Device Architecture. Designed for reliable deployment across consumer and enterprise scales. Ensures low latency, high stability, and strict data governance without compromising privacy.
Deployment & Architecture
Built for real-world deployment, not just demos. Our architecture is designed for scalable operations across consumer and enterprise environments. By leveraging on-device, privacy-first intelligence, ThingX enables reliable operation with low latency and high stability—ensuring seamless rollouts without compromising privacy or user trust.
Multimodal Intelligence
Understanding emotional context through multiple signals—on device, by design. ThingX leverages proprietary multimodal algorithms that integrate signals from voice patterns, vocal tone, and physiological data directly on-device. This approach captures emotional context more robustly than any single signal alone, ensuring consistent understanding even in noisy environments.
Digital Biomarkers
Decoding emotional signals through data. We extract and model proprietary digital biomarkers from everyday physiological signals, powered by our in-house emotion algorithms. This transforms subtle emotional states into structured, interpretable insights—enabling long-term understanding without subjective labels or diagnoses.
ThingX by the Numbers
Years Experience
15+
True innovation requires rigorous science. Our R&D is driven by an in-house interdisciplinary team comprising over 6 PhDs specializing in neuroscience, signal processing, and emotional computing, keeping us at the cutting edge of technology.
Hours Validated
30k+
Reliability is our core. Our algorithms haven't just been written; they've been tested. We have conducted over 30,000 hours of rigorous clinical and laboratory validation to ensure industrial-grade stability and precision in real-world environments.
PhD Experts
6+
True innovation requires rigorous science. Our R&D is driven by an in-house interdisciplinary team comprising over 6 PhDs specializing in neuroscience, signal processing, and emotional computing, keeping us at the cutting edge of technology.
Analyze signals
2M+
Intelligence is born from data. We have processed and analyzed over 2 million distinctive physiological data points. This massive dataset trains our models to recognize even the subtlest emotional nuances with unprecedented granularity.
Model Accuracy
97%+
Precision you can trust. Through continuous iteration and deep learning optimization, our proprietary emotion recognition models have achieved a market-leading 97% accuracy rate in controlled benchmarks, setting a new standard for the industry.
Our Technology
Incubated and Supported by the Hong Kong Science Park (HKSTP)
Operating at the intersection of innovation and industry, our strategic partnership with HKSTP empowers us to bridge the critical gap between advanced laboratory algorithms and scalable, commercial-grade solutions.