Dexcom has rolled out new AI-enabled features for its over-the-counter Stelo Glucose Biosensor System and announced plans to expand the device beyond the United States, with international launches slated to begin later in 2026 and continue through 2027 in the United Kingdom, Australia, New Zealand, and South Korea. The move extends continuous glucose monitoring, once a niche tool for insulin-dependent diabetics, toward a much broader audience of people simply curious about how food, sleep, and exercise affect their metabolism. Dexcom, founded in 1999 and now with more than 25 years in glucose biosensing, markets Stelo as the first over-the-counter glucose biosensor cleared by the FDA, a distinction it is leaning on heavily as competitors crowd into the consumer CGM space.
From Prescription Device to Wellness Wearable
Stelo made headlines when it became available without a prescription, part of a wave of CGMs marketed directly to consumers who do not have diagnosed diabetes. The device attaches to the back of the upper arm and streams glucose readings to a smartphone app, a category that has ballooned in the wellness and biohacking world even as clinical evidence for its benefit in non-diabetic, metabolically healthy people remains thin. A study published in August 2026 found continuous glucose monitors show no proven benefits for healthy people, a finding that sits uneasily alongside the surging consumer demand Dexcom and rivals like Abbott’s Lingo are capitalizing on.
What the New AI Layer Adds
Rather than simply displaying a glucose number, Dexcom’s updated AI features are designed to translate raw readings into personalized, plain-language coaching, flagging which specific meals, workouts, or sleep patterns correlate with a given user’s glucose spikes and dips over time. The redesigned Daily Insights feature uses a card-based system that surfaces up to three personalized recommendations drawn from the previous day’s glucose, activity, nutrition, and sleep data, plus a fourth card focused on behavioral coaching. Users can log meals by photographing them, scanning a barcode, or typing a text search, pulling nutritional breakdowns of calories, carbs, protein, fat, and fiber from a database of more than one million food options. Girish Naganathan, Dexcom’s executive vice president and chief technology officer, said that when the company launched Smart Food Logging the prior summer, the goal was to give users a simple and effective way to consistently log meals, and that the majority of users reported positive dietary changes within 30 days of using Stelo. That mirrors a broader industry trend toward what the FDA’s TEMPO pilot program is exploring: repositioning CGMs from purely therapeutic tools for insulin users into diagnostic and preventive screening instruments for the far larger prediabetic population, tens of millions of Americans who don’t yet carry a diabetes diagnosis but are on a trajectory toward one.
An NFL Alum Fronts the Pitch
Dexcom is leaning on a recognizable face to sell the redesigned product to mainstream consumers: Mike Golic Sr., the former NFL defensive lineman turned sports broadcaster, has signed on as a Stelo ambassador. Golic, who lives with type 2 diabetes, said the condition has forced him to slow down and pay attention to his health, and that before using Stelo he was essentially guessing about how his choices affected his blood sugar. His endorsement is aimed squarely at the millions of Americans managing type 2 diabetes or prediabetes who may never have used a CGM, rather than the smaller population of biohackers who made Stelo a wellness-culture fixture at launch.
Following FDA Clearance for Kids
The AI rollout comes on the heels of FDA approval for Stelo’s pediatric use, opening the device to a younger population and setting up potential future features aimed at helping parents and pediatricians spot early metabolic warning signs well before a type 2 diabetes diagnosis would typically occur. Diabetes technology specialists say pediatric expansion is likely to be one of the more clinically meaningful uses of the platform, given how much earlier intervention can change long-term outcomes for children showing early insulin resistance.
The Skeptical Counterpoint
Not every endocrinologist is convinced consumer CGMs, AI-enhanced or not, represent good use of healthcare attention and dollars. Some worry that healthy users chasing perfectly flat glucose curves may develop unnecessary anxiety about normal, harmless blood sugar fluctuations that occur after virtually any meal, essentially manufacturing a health concern that didn’t previously exist. Others note that AI coaching layered onto ambiguous data can generate confident-sounding advice that outpaces what the underlying science can actually support, especially for a metabolically healthy person whose glucose readings will naturally vary within normal limits regardless of what they eat.
Where the Category Goes From Here
Dexcom’s international expansion signals confidence that regulatory and reimbursement environments abroad are ready to follow the US consumer CGM boom, even as the clinical evidence base for non-diabetic use continues to lag commercial momentum. The fully redesigned Stelo app, which folds in the new AI coaching and meal-logging features, is scheduled to roll out nationwide in the US in the coming weeks before the broader international launches begin. The company’s bigger long-term bet is that AI coaching features will differentiate Stelo from cheaper CGM competitors and justify a subscription-style pricing model, while regulators decide whether to formally expand CGM’s approved use case to prediabetes screening, a decision that could reshape reimbursement and drive adoption far beyond today’s early-adopter wellness crowd. Competitors are moving in parallel: Abbott has continued expanding its own Lingo consumer CGM line with comparable AI-driven coaching features, setting up a head-to-head race between the two largest CGM makers to define what counts as trustworthy, clinically meaningful AI guidance in a product category that, until very recently, simply displayed a number and left interpretation entirely to the user.
Photo: Google DeepMind / PEXELS via Pexels
