Abbott unveils Libre Assist AI for glucose impact prediction

A recent update to the Libre 3 app in the US now enables users to snap their meal to learn how their glucose will change. At the same time the app has made it possible to see levels on a smartwatch.

Image courtesy of Abbott Diabetes Care Inc.

Users of the Libre 3 app in the US now have access to Libre Assist which uses generative AI to help PWD understand how foods may affect their glucose levels before they eat - a point which Abbott are promoting to encourage use of the app.

Built directly into the updated version of the app, the new feature allows users to:

• Snap a photo of their meal or enter a text description of foods;

• AI then predicts the meal’s potential glucose impact, displaying a colour-coded rating system —

  • green for minor impact,

  • yellow for moderate,

  • and orange for significant potential impact;

• Receive personalised guidance on how to reduce glucose spikes — from swapping ingredients to adjusting portions; and

• See personalised glucose feedback after eating, as Libre Assist connects to a FreeStyle Libre 3 Plus sensor.

Integration

After a meal, users can review actual glucose responses captured by their sensor and learn how their body reacted in comparison with the pre-meal prediction. This feedback loop reinforces personalised insight over time, and according to Abbott, will help build confidence around food choices.

Cost

Libre Assist forms part of the updated Libre 3 app so there is no additional cost to use it. There are no separate subscriptions, additional purchases, or prescriptions required either.

User reviews

Early online comments have mainly been positive, however some users point out that it is something for "fun"; that they'll tire of the novelty; that they still need to carb count and it's not something they want to keep up as the information can be got elsewhere, for example using MyFitnessPal.

Libre Assist vs MyFitnessPal

Out of interest, we've compared Libre Assist with the most popular app, MyFitnessPal below.

	Libre Assist	MyFitnessPal
Purpose	Predictive glucose guidance before meals using generative AI integrated into Libre app with CGM data support	General nutrition logging (calories, carbs, macros, diet goals)
Workflow	Photo/text meal input → AI predicts potential glucose impact and suggests adjustments	Manual entry - search, use barcodes, recipes
Pre-meal guidance	Designed for glucose impact prediction	use for planning a meal
Output	Glucose impact coded as green for minor impact, yellow for and orange with tips	Provides calories, carbs, macros
CGM integration	Integrates with FreeStyle Libre 3 sensors	None
Personalisation Note: this is not the same as individual guidance or recommendations	Uses CGM feedback to refine understanding over time	Based on user logs/goals
Cost	Free in Libre app	Free + Premium at cost
Best for	Rapid glucose-impact decisions and learning from real glucose responses	Macro & carb tracking, long-term diet goals

Limitations of Libre Assist

While Libre Assist offers proactive glucose information, there are limitations to its use in terms of measurement and its use for decision-making i.e. how much insulin to inject, which is usually calculated by a PWD on the basis of carb counting.

Below we consider some of these limitations:

1. AI predictions are not perfect

Libre Assist uses generative AI to estimate glucose impact from a meal photo or description, but:

• Generative AI may not always identify ingredients or portions accurately.

• Predicted glucose impact scores can differ from actual post-meal glucose responses.

• The feature explicitly states it should not be used to make insulin dosing or medical treatment decisions.

  
  

2. Carbohydrates are only part of the story

Glucose levels are influenced by far more than just carb quantity. It's thought there are up to 42 factors affecting glucose levels, such as:

• Meal composition (fat, protein, fibre, glycaemic index/load)

• Timing of meals and portion sizes

• Physical activity / exercise

• Stress and hormonal fluctuations

• Medications

• Hydration

• Individual metabolic traits, microbiome variation, and body state

  
  

This complexity means that even with advanced AI, any prediction inherently simplifies a very complicated physiology.

3. Individual variability is huge

As many PWD know, two people eating the same meal can have very different glucose responses and even eating the same meal can be different on different days or times.

4. Doesn’t replace clinical guidance

Libre Assist is a helpful decision support tool for meal planning, but it:

• Is not a medical device for treatment decisions (e.g., insulin dosing).

• Should supplement, not replace, professional advice from healthcare teams.

• Works best alongside real CGM trends and individual insights, not instead of them.

Libre Assist's competitors

Historically, diabetes tools have focused on carbohydrate counting, but research has shown this to be an oversimplification. Carbohydrates do matter, but they are only one of many variables, with possibly 42+ factors impacting blood glucose levels. On top of this individual glycaemic responses to the same meal can vary widely, underscoring the limits of carb counting alone.

Abbott's 'selling point' for Libre Assist is that it is predicting glucose levels before food is eaten. However, glucose monitoring apps are not new in terms of carb counting and these also use AI as well as more functionality for diabetes management including some that connect to CGM. Alternatives include. These include Snaq, Mysugr, Glucose Buddy, Diabetes:M and Carbs & Cals.

Smartwatch integration

Abbott recently updated the Libre 3 app (for use in the USA) to enable PWD using an Apple smartwatch series 4 and later to see their glucose levels directly on their watch, although the phone must still be within reach.

Interested in keeping up-to-date?

We'll be continuing to all FreeStyle Libre related news. If you'd like to be in the loop, ensure you subscribe to receive our free and independently edited newsletter, especially for CGM and pump users like you!

Click here to subscribe.

Disclaimer

Blogs and publications on this website are independent of any involvement by medtech companies or diabetes related charities. To ensure there is no bias, we do not accept any products, freebies or other material from any medtech provider. Except where credited otherwise, all materials are copyright ©️What CGM.

What CGM is not associated or affiliated with Abbott or FreeStyle Libre. Content here and on our website www.whatcgm.com does not constitute medical advice or replace the relationship between you and healthcare professionals nor the advice you receive from them.

The author of this blog has type 1 diabetes and uses a CGM on NHS prescription.