News Headlines Now

Madrid – 14 September, 2026 – Anyone living with diabetes and wearing a continuous glucose sensor has a problem that did not exist ten years ago: too much data and too little interpretation. Sensors such as FreeStyle Libre deliver one value after another, day and night, but the question that really matters, why did it go up?, is not in any of them. That is the gap GlucoBeat sets out to fill: a glucose diary in Spanish and English that combines two ideas: measuring the effect of everything you do on your glucose, and putting artificial intelligence to work reading your entire history to find what repeats.

The user logs what they do (a meal, a walk, a dose of insulin, a pill, hours of sleep) and the app takes care of the rest: it follows the glucose curve over the following hours and returns the effect in numbers. How much it rose, how fast, when it came back to the starting point and whether it left the target range. And once enough history has built up, the AI takes the next step: what agrees with you, what doesn’t, and what would happen if you ate this now.

A sensor that says how much, but not why

Continuous glucose monitoring changed the lives of millions of people with diabetes: no more finger pricks every few hours, and the ability to glance at a phone and know. But that convenience brought a silent side effect: understanding glucose is still the patient’s job, usually with no tool other than memory. The curve rises after eating, and knowing how much it rose does not tell you whether it was the pasta, the bread, that morning’s stress or the late dinner the night before.

From a patient’s frustration to a CTO’s method

Behind GlucoBeat there is no funding round and no product committee: there is a technology entrepreneur with many miles on the clock and several exits, Diego Manuel Béjar. He currently works as an external CTO (“CTO as a Service”, in industry jargon) for several startups, helping them design and build their products. And he has diabetes.

That dual condition explains the project better than any corporate deck. Someone who has spent years watching other people’s projects being born in an incubator knows how to recognise a problem worth turning into a product; someone living with diabetes has it in front of them several times a day. For months, Béjar wrestled with the basics: he wanted to zoom in on a spike in the sensor app’s chart to see it up close, jump to a specific day three weeks ago without stepping through one day at a time, write in his own words what he had for dinner without weighing anything or counting carbs, and see it in its place on the chart. Above all, he wanted to understand the effect of what he was doing.

“My job is turning problems into solutions,” he explains. “One day I asked myself the obvious question: if a startup asked me for this, how would I solve it? And I started building it.”

What happens to your glucose after everything you do

The heart of the app is a calculation that repeats with every event logged. GlucoBeat takes the glucose level of the previous half hour as a baseline and follows the curve over the hours that follow: three hours for a meal, four for exercise or insulin, eight for sleep. It records how far glucose rose or fell, how many minutes it took to get there, how long it took to return to the starting level and whether at any point it left the target range. When two things that move glucose overlap, such as a meal and the walk afterwards, the app flags it rather than attributing the effect to whichever came first.

None of that replaces the basics: GlucoBeat also tells you how much glucose you have. Readings come in by importing the full FreeStyle Libre history from LibreView, by connecting LibreLinkUp so they arrive on their own every few minutes, or by hand, from any other meter. The current value is shown on the dashboard, in the day view and even on the watch. The difference is what comes after that number: the how much gets a why.

So the app shows the latest meals and what they did to the person who ate them: what was eaten, how much glucose rose and what the curve looked like over the following three hours. The statistics gather what gets looked at in the endocrinology clinic (time in range, variability, estimated glycated haemoglobin, hypoglycaemia and the profile of a typical day), plus the average impact of each type of event and each food, compared with the previous period. The app adapts to each profile: type 1 diabetes, type 2, other types, or simple tracking without diabetes.

“Your instruction manual is in there”

The project’s second engine is not diabetes: it is the trade. Béjar works with artificial intelligence every day, it is a large part of what he does for the companies he advises, and the question posed itself. “If I use it to get value out of other people’s data, how could I not use it on my own?”

The reasoning that follows underpins the most ambitious part of GlucoBeat. A glucose history is not a handful of pretty charts. It is months of readings taken practically minute by minute (a sensor that records a value every minute exceeds half a million glucose readings a year), cross referenced with everything its owner has eaten, walked, slept and injected. It is the most detailed record in existence of how a body with diabetes responds to daily life. And it almost always goes to waste: no human eye can read through all of it, let alone find what repeats.

“Your history is not a pretty chart: it’s months of minute-by-minute readings crossed with everything you’ve eaten, walked, slept and injected. Your instruction manual is in there, and no human eye can read it all. AI can,” says Diego Manuel Béjar.

What artificial intelligence finds in your glucose

That is where the AI analysis comes from. The user picks a period (the last week, a month or a quarter) and the AI goes through all their glucose readings and everything they have logged to tell them what repeats. It does not return averages but concrete findings, each with its figure and the number of times it has happened.

It looks, for instance, at which meals raise glucose the most and which barely move it; how exercise affects it, comparing only sessions of the same type and intensity (because a walk and a weights session can move glucose in opposite directions) and also looking at the following night and whether lows appear in the next 24 hours; at what times of day it tends to drop and what usually happens beforehand; whether there are dawn rises, differences between weekdays and weekends, or changes after a bad night’s sleep. On the app’s website, a sample analysis concludes that “your dinners raise you quite a bit more than your lunches”, with findings such as a pizza at night that raises glucose by an average of 92 mg/dL across four dinners, or a walk after lunch that leaves the peak 35 mg/dL lower.

Each analysis ends with something designed for the clinic: specific questions the patient can take to their doctor or diabetes educator. And it says so when the data is not enough to draw conclusions, for example when there are days in the period with no entries, rather than inventing patterns. The AI always speaks of correlations in that person’s data, never of causes, and adapts its language: it does not speak the same way to someone with type 1 diabetes as to someone with type 2 not on insulin, or to someone who simply wants to understand their metabolism.

“Can I eat this?”: AI before the first bite

The same logic, in pocket format, gives its name to the app’s most direct feature. The user types what they are thinking of eating (the more detail the better: quantities, sides, drink) and the AI tells them whether it suits them, at this time of day and with the glucose they have right now.

To answer, it starts from their current glucose and trend, takes into account what they have eaten or the exercise they have done in the last few hours and the time of day (the same meal does not land the same way at noon as at night) and, above all, searches their history for similar meals and how they went. With all that it returns the predicted glucose peak, how long it would take to get there and the estimated carbohydrates in the dish, plus practical ideas to make the rise smaller: a flat plate instead of a bowl, starting with a salad, saving the fruit for a snack or going for a walk afterwards, if their own data says that works for them.

In the website’s example, someone about to eat a bowl of macaroni with tomato and cheese and an apple, at 132 mg/dL and rising, gets a predicted peak of 205 mg/dL after an hour, around 95 grams of carbohydrates and a warning that the dish would take them above their range for a good while. The answer explains what it was based on, how confident it is and what would be worth detailing next time. What it never does is say “you can’t eat this”: it describes what will probably happen and leaves the decision in the hands of the person asking.

Designed for everyday life with diabetes

GlucoBeat’s most recent features target the hardest part of any diary: not giving up on logging. The app works offline for what matters: in a restaurant with no signal you can log the meal, which is saved on the phone and uploaded on its own as soon as the network is back.

It also includes medication reminders, daily, weekly, fortnightly or monthly, at the time the user chooses; tapping the notification opens the app with the entry already prepared, and all that is left is to save it. And a daily check-in: a notification at the chosen time which, if nothing has been logged that day, reminds you, and if something has, invites you to look at the effect it had. No notification carries glucose data, because they are read on the lock screen.

All of this in an app that runs from the browser, with no downloads, and is added to the phone’s home screen like any other app. It has a button for logging always at hand, with the time already filled in.

Tools for doctors to follow their patients

GlucoBeat also addresses healthcare professionals, with a simple proposition: see the patient’s real diary between appointments, not what they remember on the day of the visit. The doctor creates a free companion account, which asks for no data of their own, and it is the patient who shares their diary from the settings and decides for how long: a week, a month, three months, a year or open-ended.

From then on, the professional has each patient in their following list and can look at their full day, with the sensor curve and what was logged on top of it, their time in range and statistics, the measured effect of every meal, every walk and every dose, and their medication with the exact time. If the patient’s sensor is connected, readings arrive in real time. The doctor can also keep their own notes on each patient, visible only to them.

The approach looks after the patient: access is read-only, the doctor cannot change anything in the diary, the patient withdraws it whenever they want, and every look at their data is logged and visible to the patient, with date and time. The professional account is free and has no patient limit.

The same system serves families and caregivers, for example parents of children with diabetes. The app explains the whole process on a dedicated page for doctors.

AI, with the brakes on

The enthusiasm for artificial intelligence comes with a counterweight, and it is the most unusual part of the project. The AI only receives an anonymous summary of the data (no name, no email and no real dates) and only when the user asks, under a specific consent that can be withdrawn at any time. Each analysis stores, where the user can see it, exactly what was sent.

The rules are written so they cannot be crossed: the AI never recommends or calculates insulin or medication doses, never suggests changing a treatment and never diagnoses. Nor does it comment on anyone’s weight or body. And what it returns still goes through a second filter: any sentence that mentions an amount of insulin or proposes adjusting a treatment is removed before it reaches the user, with a notice that this has been done. Results are labelled as AI-generated, in line with the European AI Act, and come with a medical disclaimer.

Glucose data treated as health data

Data is stored encrypted on servers in the European Union, in accordance with the GDPR, and users sign in without a password, with a code sent by email. Consents are separate for each use (health, artificial intelligence, syncing and sharing) and can be withdrawn at any time. Users can download all their data in one click and permanently delete their account with email confirmation, without writing to anyone or waiting.

Free to start, paid if you want more AI

The diary, the charts, the statistics, history import, sensor syncing and sharing are free forever with no card required, and include a first taste of the AI: three analyses and five “Can I eat this?” queries.

Since AI is what costs money, AI is what separates the paid plans. Premium, at 5.95 euros a month or 59 a year, offers eight analyses and twenty queries a month, designed for someone who runs one analysis a week. Premium+, at 9.95 euros a month or 99 a year, goes up to twenty analyses and fifty queries a month for those who use the AI daily. Both can be cancelled at any time, with no lock-in, and, in the app’s own words, those who pay help keep GlucoBeat free for those who cannot afford it.

What it is not

It is worth saying as clearly as the app itself does: GlucoBeat is not a medical device. It does not diagnose, does not calculate doses, does not replace any professional and does not make decisions for anyone. It is a diary that records, visualises and shows correlations in its owner’s glucose. Nor is it affiliated with Abbott, the manufacturer of FreeStyle Libre.

Its stated ambition is more modest and, probably, more useful than that of apps promising to control diabetes: that those who live with it understand their glucose, keep it more balanced and arrive at their appointment with the information in order.

This is only the beginning

GlucoBeat is available in Spanish and English and works the same on the phone, on the computer and on the watch. Its creator describes it as a living project, growing with whatever he himself finds missing each day.

“GlucoBeat will keep growing with what we learn along the way,” he says. “Every improvement comes from the same question I asked myself: what do I need to understand myself a little better? If you live with diabetes, or care for someone who does, you are welcome. This app is yours too.”

About GlucoBeat

GlucoBeat is an AI-powered glucose diary, in Spanish and English, for people with type 1, type 2 or other types of diabetes and for anyone who wants to understand their metabolism. It links sensor or meter readings with meals, exercise, insulin, medication and sleep, measures the effect of each on glucose, and analyses the history with AI to find patterns and predict how a meal will land. It lets doctors, family members and caregivers follow the patient’s diary with read-only access. It is an independent project developed in Spain, with data hosted in the European Union. It is not a medical device: it does not diagnose or recommend doses.

Press contact

Diego Manuel Béjar, founder of GlucoBeat

Email: [email protected]

Screenshots of the app and logo available on request.

Media Contact
Company Name: GlucoBeat
Contact Person: Diego Manuel Béjar
Email: Send Email
Country: Spain
Website: https://www.glucobeat.com