KOS AI, a Silicon Valley health technology firm, has developed a non-invasive glucose monitoring device that could potentially transform diabetes management for millions worldwide. The company has appointed Stanford CS graduate Abhinav Agarwal as Chief Technology Officer, following his pivotal role as Head of Machine Learning and AI in developing the core technology behind the KOS Argus device. Agarwal will now lead the continued expansion of the company’s AI-powered health monitoring platform.
The KOS Argus device uses optical sensing technology combined with machine learning algorithms developed by Agarwal’s team to measure blood glucose levels through the skin without requiring needle pricks or invasive sensors. According to the company, the technology achieves over 91% accuracy in glucose monitoring and demonstrates 93.5% sensitivity in detecting potentially dangerous low blood sugar events.
“This technology represents a significant advancement in preventive healthcare through AI application,” said Abhinav Agarwal, newly appointed CTO at KOS AI. “We developed algorithms that allow the system to adapt to each user’s unique physiological patterns, potentially improving accuracy over time while maintaining data privacy by processing information locally on the device.”
The innovation comes at a critical time as the global diabetes monitoring market continues to expand, with an estimated 537 million adults worldwide living with diabetes. Traditional glucose monitoring methods often involve discomfort that can lead to reduced testing frequency and potentially compromised health outcomes.
Under Agarwal’s leadership, the machine learning division at KOS AI pioneered an adaptive AI system that forms the technological foundation of the Argus device. His team’s work on efficient AI algorithms enables the device to perform complex calculations without excessive battery drain, a critical feature for wearable health technology.
Casey Nguyen, Head of Product Strategy at KOS AI, highlighted the broader implications: “Beyond glucose monitoring, the optical sensing and AI technologies we developed have applications across multiple health metrics, potentially transforming how we approach preventive healthcare.”
From left to right, Casey Nguyen, Abhinav Agarwal, and Schalal Habib
KOS AI presented additional details about the technology during a press conference held on March 20, 2025, at their Palo Alto headquarters. The company conducted a live demonstration comparing the KOS Argus to current market solutions, highlighting its non-invasive approach against traditional finger-prick methods.
During the demonstration, KOS AI revealed significant inconsistencies in traditional glucose monitoring devices, showing that samples drawn from the same finger at the same time produced 30-40% variance in readings—a fluctuation that is physiologically impossible. This comparison emphasized the potential advantages of the AI-driven optical sensing technology developed under Agarwal’s direction in providing more consistent measurements without requiring blood samples or needles.
The company is currently conducting clinical validation studies and anticipates FDA clearance by late 2025. At $349, the device is positioned as a more affordable alternative to traditional continuous glucose monitoring systems that can cost thousands annually when accounting for replacement sensors.
KOS AI was founded by Schalal Habib, who holds five patents with 45 more pending approval. Agarwal joined the company after completing his Stanford CS degree, where his and Casey Nguyen’s research on “Structured Pruning for Large Language Models” was recognized at ICAIRME 2025. His expertise in machine learning efficiency directly contributed to the Argus device’s ability to run sophisticated algorithms on a wearable form factor.
About KOS AI
KOS AI is a health technology company based in Palo Alto’s Stanford Research Park, developing AI-powered, non-invasive health monitoring solutions. The company combines optical sensing technology with advanced artificial intelligence to create accessible and comfortable health monitoring devices. KOS AI focuses on creating platforms that provide comprehensive health insights while maintaining user privacy and comfort.