NutriSight uses live camera AI, your personal health goal and meal memory to help you choose better — at cafés, supermarkets, on your phone and on future AI glasses.
An Education & Human Potential product: we teach people what food actually does to their body — at the moment of decision.
“Help every human on Earth understand what food does to their body — before they eat it.”
Why glucose is worth watching
Most cancer cells take up far more glucose than healthy tissue — the Warburg effect. It is why PET scans use a radioactive sugar tracer to find tumours. That biology is well established, and it is one reason understanding your carbohydrate load is worth learning. What follows is education, not treatment advice.
Read this before changing anything
NutriSight is educational and does not diagnose or treat disease. A ketogenic or low-carb diet is a supplement to — never a replacement for — standard cancer treatment. Always talk to your doctor or oncologist before changing how you eat during cancer treatment, pregnancy, type 1 diabetes, or with a history of eating disorders.
In NutriSight you can set Ketogenic / low-carb as your focus goal — every scan is then scored on net carbs and sugar load.
The problem
NutriSight helps you decide before you eat — with vision, personal goals and learning.
You can't see sugar, fat, salt and portions hiding in everyday meals.
Manual logging is annoying. The choice was already made hours ago.
Real food decisions happen at cafés, supermarkets and at home — not later in an app.
Education
Nutrition labels can be hard to understand in the moment. Calories, grams of sugar, carbohydrates, salt, protein and fiber are useful numbers, but many people do not know what they mean when they are about to choose food.
Glucose science
Some foods do not taste sweet, but can still have a strong effect on blood glucose. Foods such as rice, bread, pasta and potatoes contain starch, which is a type of carbohydrate. During digestion, many carbohydrates are broken down into glucose, which enters the bloodstream and can raise blood glucose levels.
Glucose is a sugar in the blood and an important energy source for the body. After eating carbohydrates, the body breaks many of them down into glucose. This is normal, but large or frequent glucose spikes may matter for people who want stable energy, weight management, or better blood-sugar awareness.
Visual comparison
These simplified examples show how different foods can affect glucose and energy differently.
Fast energy, high sugar, lower fiber
~6 sugar cubes worth of glucose effect
May not taste sweet, but starch can break down into glucose
High blood-glucose effect
Depends on dressing, toppings and portion size
Lower blood-glucose effect
These are simplified estimates for education. Real nutrition depends on portion size, ingredients, preparation and the individual.
Low-carb & keto
When carbohydrate intake stays very low, the liver makes ketones from fat instead of relying on glucose. NutriSight now includes a Ketogenic / low-carb focus goal that scores every scan on net carbs and sugar load.
Educational information only. Always discuss diet changes with a qualified clinician — especially during illness, cancer treatment, pregnancy or when taking medication.
Dr. David Unwin / Public Health Collaboration sugar infographics.
phcuk.org/resourcesImportant note
NutriSight provides estimated nutrition guidance and is not medical advice. Nutrition estimates may be inaccurate. People with diabetes, eating disorders, pregnancy, medical conditions or special dietary needs should consult a qualified healthcare professional.
vs traditional trackers
Most apps log food after the damage is done. NutriSight is built to help you decide before you eat.
| Capability | NutriSight | Traditional calorie trackers |
|---|---|---|
| Scan food from a photo | Yes | No |
| Read menus and recipes with vision AI | Yes | No |
| Parse meals from plain text | Yes | Sometimes |
| Compare two food choices side-by-side | Yes | No |
| Goal-based recommendation per scan | Yes | No |
| Manual barcode logging | Optional | Yes |
| Calm, non-judgmental tone | Yes | Mixed |
| Future-ready for AI glasses | Yes | No |
Three ways to log a meal
| Mode | Best for | Speed | Accuracy |
|---|---|---|---|
| Scan | Plates, dishes, café meals | ~3 seconds | Medium (vision-based) |
| Recipe | Printed recipes with gram amounts | ~5 seconds | High (reads exact grams) |
| Menu | Restaurant or café menus | ~5 seconds | Medium (picks best match) |
| Describe | Quick text or voice logging | ~4 seconds | Medium (depends on detail) |
| Compare | Choosing between two options | ~6 seconds | Medium (relative is easier) |
Honest scope
Quick answers
Designed to be readable for humans first — and extractable by Google AI Mode, AI Overviews and ChatGPT-style answer engines.
NutriSight is an AI nutrition assistant that helps people understand food before and after eating. It scans meal photos, reads menus and recipes, parses meals from plain text, and compares two food choices side-by-side — then gives a calm, goal-based suggestion. Estimates only, never medical advice.
Open the camera or upload a photo. NutriSight's vision model identifies the dish, estimates a typical serving, and returns calories, protein, sugar, carbs, fiber and salt with a confidence level. You can edit any value before saving it to your daily meal memory.
Yes — NutriSight estimates calories, macronutrients and a goal-based health score from a single photo. AI vision is approximate, so values are shown as estimates with low, medium or high confidence. For recipes with printed gram quantities, accuracy is much higher.
Yes. The Choose Better feature accepts two food photos and returns a side-by-side table for calories, protein, sugar, fiber, satiety and blood-sugar stability — plus a one-line recommendation for your goal. Useful at cafés, restaurants and supermarket shelves.
Not in the traditional sense. NutriSight focuses on decisions before you eat, not after. It still keeps a daily meal memory and totals, but the goal is fewer manual entries, more visual understanding, and one helpful next-best-action — instead of obsessive logging.
No. NutriSight provides educational nutrition estimates designed to support healthier choices. It does not diagnose, treat or replace professional medical advice. People with diabetes, eating disorders or other medical conditions should always consult a qualified clinician.
AI food estimates depend on visible portions and lighting. Expect ±20–30% error for typical meals, and tighter accuracy for recipes with printed grams or for packaged items. NutriSight always shows a confidence level so you know when to double-check.
NutriSight highlights carb impact and sugar load and suggests pairings that flatten glucose response — protein, fiber, healthy fats. It is decision-support, not a glucose monitor, and does not replace medical care for people with diabetes.
Anyone who wants to make better food choices without manually logging every bite. It is especially useful for people targeting weight loss, more protein, less sugar, more fiber, stable energy or healthier eating overall — at cafés, supermarkets and at home.
Hidden ingredients, oils and dressings are hard to estimate from a photo. Heavily mixed dishes lower confidence. NutriSight is not a medical device and does not measure glucose, ketones or micronutrients. Always edit estimates if you know better numbers.
The solution
Less sugar, more protein, stable glucose, lose weight, etc.
Point your camera. Tap. AI sees food, menus, café counters.
Calories, sugar cubes, carbs, protein, fiber, salt — all visible.
Personalized to your goal and what you ate today.
Education + gamification
NutriSight isn't just a scanner — it's a coach. Each focus goal unlocks short lessons that teach the why behind the numbers. Finish lessons to earn XP, build streaks, and turn understanding into habit.
What sugar actually does to body fat. Why protein is the cheat-code macro. How to flatten glucose spikes. Real science, 2-minute reads.
Every lesson and every scan gives you XP. See your knowledge grow alongside your healthier habits.
Open the app once a day, scan one meal. The streak builds. The habit builds itself.
Visual intelligence
Numbers don't change behavior. Pictures do. NutriSight turns invisible sugar into visible cubes.
Soda 330ml
Bowl of white rice
Rice doesn't carry sugar cubes — we show its blood-sugar impact instead.
Under the hood
Identifies dishes, ingredients and portions from the camera.
Reads café and restaurant menus to extract real choices.
Estimates calories, sugar, carbs, protein, fiber and salt.
Remembers your day to keep guidance personal, not generic.
Teaches the science behind every recommendation.
Adds disclaimers and avoids medical claims.
—
Users
—
Scans analyzed
—
Lessons completed
—
Beta signups
"Use this slot for a real beta tester quote."
— Beta tester slot
"Use this slot for a real beta tester quote."
— Beta tester slot
"Use this slot for a real beta tester quote."
— Beta tester slot
Placeholder slots — to be filled with real verified quotes only. We never invent testimonials.
Built with
Google Gemini 2.5 Flash & 2.5 Pro — multimodal vision and reasoning for every scan, menu read, recipe parse and comparison.
Google Cloud-powered AI pipeline via the Lovable AI Gateway, with managed Postgres & Auth for secure meal memory and storage for food images.
See every agent and live counter on the AI Operations page.
Create a free account to save scans, build streaks and earn XP as you learn what food really does to your body.
Beta access
We're onboarding founders, dietitians, students and curious eaters. Tell us your goal and we'll be in touch.