From Stalled to Standard-Setting: Building the World's First Commercially Viable Drug Impairment Detection Platform for Gaize

Gaize set out to solve a problem that traditional drug testing can't touch: detecting whether someone is impaired right now, not whether they used a substance days or weeks ago. The science was sound — automated versions of the Drug Recognition Expert (DRE) ocular tests that law enforcement has relied on for over 45 years — but the technology to turn raw eye-tracking signals into trustworthy, defensible results didn't yet exist.

Before bringing us in, Gaize had invested in outside contractors to build that capability. The engagement left them empty-handed: no working detection pipeline, no reliable signal extraction, and no clear path from their VR-captured eye data to an impairment result a safety director — or a court — could stand behind.

The core difficulty wasn't a lack of data. It was the opposite. Each test produces an enormous volume of high-frequency ocular measurements, and the signals that distinguish impairment from normal variation are subtle, noisy, and easy to miss. Gaize needed a team that could find those signals and engineer them into a production-grade system.

We started where the previous effort fell short: in the data. We ran an extensive data-analysis and data-science program across Gaize's eye-tracking dataset — one of the largest of its kind, drawn from the world's largest cannabis impairment clinical trial and spanning hundreds of millions of data points — to isolate the ocular markers that reliably indicate impairment.

From there, we rebuilt Gaize's platform from the ground up:

• Signal extraction from raw ocular data. We developed methods to pull meaningful impairment signals out of high-frequency eye-movement, pupil-response, and gaze data, separating real indicators from noise.
• Sophisticated machine learning models. We replaced ad-hoc logic with rigorously developed ML models capable of detecting signs of impairment from both alcohol and cannabis.
• A re-architected, production-grade platform. We designed the architecture to run automated DRE-based tests, analyze the results in real time, and deliver clear, defensible impairment reports — repeatably and at scale.

The result was a system that doesn't just process data, but turns it into evidence safety leaders can act on with confidence.

Our work took Gaize from a stalled prototype to a market-defining product. We enabled high-quality detection of both alcohol and cannabis impairment, and in doing so helped Gaize become the world's first commercially viable drug impairment detection technology — now trusted by industrial leaders across construction, mining, manufacturing, and logistics.