Abbott Just Paired a Glucose Monitor With a Dietitian. Mr Bite's Health Sync Has Never Once Asked a Phone for That Number.

Abbott and Berry Street announced an exclusive partnership on September 23rd that pairs Lingo, Abbott’s over-the-counter continuous glucose monitor, with live coaching from Berry Street’s registered dietitians. The pitch is specific: Berry Street’s RDs get trained on reading CGM data, Lingo and Berry Street users get matched with one of them, and the dietitian turns a glucose trend line into “informed” meal-by-meal advice instead of a number nobody on the user’s end knows how to act on. Berry Street users get Lingo at a 15 percent discount to make the pairing easy to start. It’s not an isolated move. Dexcom spent most of this year rebuilding its own Stelo app around the same idea, adding AI meal-photo logging that projects a food’s glycemic impact before you even eat it. Two separate companies are converging on the same product shape: a sensor that measures what your body actually does with food, and a coach, human or AI, sitting on top of it to make that number legible.

That shape is close enough to what Mr Bite already claims to do that I wanted to check our own version of it. We already sync Apple Health and Google Health Connect, and the coach already reads that synced data back to you: steps, heart rate, sleep, calories burned. Blood sugar is the one number that most directly answers the question this whole app exists to help with, what a given meal actually did to you, so I went looking for where it lives in our health sync. It doesn’t.

The HealthKit permissions block in healthService.ts reads Steps, StepCount, ActiveEnergyBurned, BasalEnergyBurned, RestingHeartRate, SleepAnalysis, DistanceWalkingRunning. No BloodGlucose, despite it being a standard HealthKit quantity type that any Lingo, Libre or Dexcom sensor writes into automatically the moment a user pairs one. Android’s Health Connect integration is narrower still: a single record type, ActiveCaloriesBurned, and the comment above it explains why on purpose rather than by accident, Google Play rejects health permissions that don’t map to a visible feature, so even step count and heart rate were left out there, let alone glucose. Bite’s health sync isn’t an oversight that happened to skip blood sugar. It’s a deliberately scoped list, and glucose was never on it.

Here’s the part that made me actually write this issue instead of shrugging. The word “glucose” is already in our code, twice, in the regex lists that decide how Mr Bite’s chat coach handles a question: HEALTH_QA_TOPIC_RE in mr-bite-chat/index.ts and its twin in mrBiteChatLoopPolicy.ts, both used to detect when someone’s asking a real health question so the model answers carefully instead of casually. Sitting right next to that logic is a separate regex, APP_CONTEXT_READ_LANGUAGE_RE, that detects when a user is asking about their own synced data: steps, activity, hydration, sleep, heart rate. Glucose never appears there, for the same reason it never appears in the HealthKit permissions list. The code already knows glucose is a careful-answer topic. It has never once been taught that glucose could be a number about you.

So if a user asks Bite “does rice spike my blood sugar,” the coach is explicitly routed to treat that as a health question worth answering carefully, and it has nothing of the user’s own to answer it with. It can only give the generic version of the answer, the same paragraph it would give anyone. Abbott and Berry Street’s whole premise this month is that the generic version is exactly what a CGM reading replaces.

I’m not pretending this is a quick fix. Reading blood glucose means a real onboarding flow, real medical-adjacent data handling, and a reason to trust a number that most of our users have never had a sensor to produce in the first place, not a one-line addition to a permissions array. But the gap is worth naming plainly instead of leaving it implied. The word is already in our code, and we’ve built real logic around treating it carefully. The number behind the word has never once been asked for.

Three Quick Bites

1. Researchers presented Phase 2 results for CBL-514, a first-in-class injection that triggers fat-cell apoptosis, at the European Association for the Study of Diabetes meeting in Milan (September 28 to October 2). The drug reduced subcutaneous fat volume by at least 150 milliliters in 69.6 percent of treated participants by week eight, versus zero percent on placebo, and cut visceral fat nearly 12 percent at week four. It’s a genuinely different mechanism from a GLP-1: not appetite suppression, direct fat-cell death. Medical Xpress

2. A dose-response meta-analysis published September 22nd in Family Medicine and Community Health pooled 51 prospective cohort studies covering 8,819,894 adults and found higher ultra-processed food intake linked to greater risk of digestive disease, high blood pressure, metabolic syndrome, and death. It’s a different, much larger paper than the UCSF depression pilot I covered two issues ago, same general direction, an order of magnitude more people. FoodNavigator

3. The FDA announced on October 2nd that Frutas y Hortalizas del Sur S.A. expanded its frozen organic berry recall again, adding lots of Great Value Triple Berry Blend, Great Value Blueberries, and Trader Joe’s Mixed Berry Blend over possible E. coli O145 contamination. It’s the third expansion since the recall first started in July, now linked to 17 illnesses and 6 hospitalizations. A recall that keeps growing three months in says more about how far shipped lots can travel than about any one bag of fruit. FDA

Tool of the Week: the University of Sydney’s Glycemic Index Database

Since this issue is about a number most people have never had a sensor to measure, this week’s pick is the closest free substitute: the University of Sydney’s searchable glycemic index and glycemic load database, built from its own peer-reviewed food testing going back decades. Search any food and get its measured GI, its GL per serving, and the study it came from, not a guess. It won’t tell you what rice actually does to your blood sugar today. It’s the best free estimate of what it does to most people, and it’s a better starting answer than the generic one our own coach is currently stuck giving. Glycemic Index Database

Thanks for reading. If you want a coach that’s honest about the number it’s never asked your phone for, Mr BITE is free on iOS and Android.

Collins, building Mr BITE