For the 537 million people living with diabetes worldwide, blood sugar monitoring has long been an inconvenient and painful necessity. Finger pricks, implantable sensors, and costly continuous glucose monitors (CGMs) dominate the market, but KOS AI has something different — Argus, a fully non-invasive, AI-driven glucose monitor that doesn’t break the skin.
Set to be unveiled on March 20, 2025, at Stanford Research Park in Palo Alto, CA, the KOS Argus could mark a fundamental shift in how we approach glucose monitoring.
AI Meets Biosensing: A New Way to Track Blood Sugar
Traditional CGMs rely on electrochemical sensors inserted under the skin. Argus, however, leverages multi-wavelength optical sensing and artificial intelligence to measure glucose levels without invasive procedures.
How does it work? The device shines light through the skin, measuring how different wavelengths interact with blood chemistry. Machine learning algorithms then process these optical signals, identifying glucose concentration with remarkable accuracy. The result:
- No finger pricks
- No implanted sensors
- Real-time tracking
- Minimal calibration
Moreover, it’s all priced at $349, making it more affordable than traditional CGMs.
This AI-driven approach enables 91.3% classification accuracy, 93.5% sensitivity for hypoglycemia detection, and 97.2% specificity — metrics that place Argus in the upper tier of glucose monitoring solutions.
The AI Behind the Breakthrough
The real innovation isn’t just the hardware, it’s the machine-learning models powering Argus. KOS AI’s proprietary structured pruning approach optimizes neural networks to analyze multi-wavelength photoplethysmography (PPG) signals in real time, reducing processing latency while maintaining high accuracy.
Abhinav Agarwal, KOS AI’s Head of Machine Learning, leads the company’s AI development team responsible for the breakthrough signal processing framework powering the Argus device. A Stanford University Computer Science graduate with a specialized focus on healthcare AI applications, Agarwal’s pioneering research with Casey Nguyen on structured LLM pruning earned Best Paper at the ICAIRME 2025. His innovations reduced model size by 35% and inference time by 40%.
Beyond his industry contributions, Agarwal is passionate about AI education, serving as an instructor at Khan Lab School and Inspirit AI, where he mentors students in machine learning fundamentals and healthcare applications. Before joining KOS AI, he developed AI applications at GDX Co. and led research projects on efficient machine learning for healthcare applications.
“This is the first time AI has been successfully integrated into a non-invasive glucose monitoring system at this scale,” says Agarwal. “Our approach bridges biomedical signal processing and deep learning, allowing for high precision with minimal user burden.”
Disrupting a Multi-Billion Dollar Market
The global continuous glucose monitoring (CGM) market is projected to grow from $13.13 billion in 2025 to $24.07 billion by 2030. The industry has been dominated by Dexcom (NASDAQ: DXCM) and Abbott Laboratories (NYSE: ABT), whose CGMs require ongoing sensor replacements, adhesive patches, and recurring supply costs.
Argus eliminates these pain points. Its one-time device purchase and low-maintenance AI-driven tracking challenge the subscription-heavy business model of existing players. If successful, this needle-free technology could pressure the industry to rethink how diabetes care is delivered.
March 20: The Moment of Truth
KOS AI will host a press conference and live demonstration of Argus at its Palo Alto headquarters. Journalists, investors, and industry leaders will get hands-on experience with the device, testing its real-time capabilities.
The event will take place at KOS HQ in Stanford Research Park (3000 El Camino Real, Building 3, Suite 120, Palo Alto, CA 94306) on Thursday, March 20, 2025, from 3:00 PM to 4:00 PM PST. Attendees will get a behind-the-scenes look at the technology, with hands-on access to its advanced features.
Media and industry professionals interested in attending can RSVP here. For additional details, visit the KOS AI website.
The Next Leap in AI-Driven Health Tech
With FDA clearance expected in Q4 2025, KOS AI is building strategic partnerships with healthcare providers and insurers to integrate Argus into mainstream diabetes care. The company is betting on a future where AI-powered wearables replace invasive, outdated monitoring methods.
The potential impact goes beyond diabetes. If AI can accurately track glucose levels non-invasively, what’s next?
KOS AI is proving that the next revolution in health tech won’t just be wearable, it’ll be invisible.
