Biteline
Guides and research
Long-form guides to calorie targets, portion estimation and logging real meals — plus our own benchmark data on how accurate AI food scanning actually is.
No listicles and no invented statistics. Each of these answers one question properly, admits where the evidence runs out, and says so.
- Original researchHow accurate are AI calorie counting apps? We tested 30 meals.Original research: 180 scans, three models, one production rubric. The models disagreed with each other by a median of 49% — and all of them passed the arithmetic check.12 min read
- Macros explainedProtein, carbs and fat: what split actually matters for weight loss?Only one of the three numbers is worth setting on purpose, and it is protein. The rules, the guard rails and two worked examples you can check.9 min read
- Calorie targetsHow many calories should I eat to lose weight?Maintenance minus a deficit, worked through with real numbers — including the case where the safety floor overrides the pace you asked for.11 min read
- Practical trackingHow to log restaurant and takeaway mealsYou cannot see the oil, weigh the rice or read the recipe. Break the plate into parts, compute the drinks exactly, and round up.11 min read
- Portion estimationHow to count calories without weighing your foodA scale answers one of the four questions a meal asks. Learn what your own hand holds, spend your precision on the oil, and eyeball the broccoli.11 min read