The AI-powered CGM boom: the glucose app from Diana Bolocco's son enters a race Dexcom and Abbott are already running

By El Ecosistema Startup · October 1, 2026.

Diego Cisternas, the second son of Chilean TV host Diana Bolocco, is developing an app that combines a continuous glucose monitoring (CGM) sensor — a patch that measures blood sugar around the clock, with no finger pricks — with an artificial intelligence model. The goal, as Cisternas himself has explained on social media and on his campaign's website, is to translate that data into concrete recommendations on diet, sleep, stress and training. He is looking to recruit 1,000 people for a closed beta; so far he has signed up around 300, according to the team itself.

The mechanism is simple on paper: the user receives the sensor at home, places it on their arm, and the app turns each reading into everyday language — which meal spiked their glucose, how a bad night's sleep affected them, what happens to their metabolism when they train or get stressed. It is also the family's second foray into the sector: his brother Pedro Cisternas already founded NutrIA, another AI nutrition startup that has grown across Latin America.

The story comes wrapped in celebrity gloss, but its timing is no accident. Dexcom, one of the big CGM manufacturers, recently presented new clinical data on its G7 sensor in adults with type 2 diabetes not using insulin, and in parallel announced an app with AI features and a planned acquisition of Nutrisense, a personalized nutrition platform, as reported by Yahoo Finance. Abbott, for its part, sealed a multi-year alliance with Google Health to connect its over-the-counter CGM Lingo with an AI assistant, Health Coach, and announced one of the world's largest metabolic health studies, combining CGM, wearables and lab testing, according to Engadget. In June, the U.S. Food and Drug Administration (FDA) approved Stelo, Dexcom's over-the-counter CGM, for use in children from age 2 who do not use insulin, according to TechRepublic, abruptly widening the potential market. The underlying point is that hardware is no longer the barrier: with over-the-counter sensors and open APIs, any small team can build the software layer on top, and that explains why projects like Cisternas's keep appearing even in a market with limited venture capital such as Chile's.

It is worth keeping in mind the caution that already exists within the sector itself. Engadget warned back in 2023 that evidence on the benefits of CGM in people without diabetes is limited, and a later analysis by the Johns Hopkins Bloomberg School of Public Health insisted that a healthy body regulates most glucose spikes on its own, so obsessing over keeping readings artificially low can induce bad habits rather than correct them. Physician Idrees Mughal, known as Dr. Idz, went as far as telling Engadget that even he, as a professional, would prefer to cross-check sensor data against other markers before making clinical decisions, and that the average user is unlikely to know how to interpret those readings on their own.

Our reading is that two things that are easily conflated should be kept apart here: media curiosity about the surname and the product's actual merit. The surname opens doors — visibility, coverage, possibly early investment — but it is no substitute for clinical evidence, and the scientific voices cited in the coverage already mark the limits of what a CGM can tell a healthy person today. It is the same pattern we see again and again in applied AI: demonstrated capability — data, studies, pilots with thousands of users — has to be distinguished from the marketing narrative. A closed beta of 1,000 people is, in that sense, the right move: it generates real signal on retention and behavior before raising capital or scaling.

Over the long term, the substance of the matter strikes us as promising and connects with a thesis we hold: the combination of cheap biometric sensors and AI models capable of interpreting them in real time is one of the most direct routes to preventive, personalized medicine — the kind that at scale can help delay metabolic disease and extend years of healthy life, not just years of life. But the short term demands honesty: without medical supervision, without making clear what the model can and cannot infer, and without clear regulatory frameworks — the FDA is already moving, Latin America is not yet — these apps risk selling certainty where there is only correlation. Latin America's digital health ecosystem is taking its first steps; how many of these projects cross the line from media anecdote to sustainable company backed by evidence is the question that really matters, and it has no answer yet.