Yes, it helps a lot. I've coded my own workout and tracking macro app with Gemini label scanning, which works wonders. A 3500 calorie deficit is equivalent to one pound lost. I set a goal to hit that and I'm halfway there. It really helps at the end of the day to check where I am on my smartwatch and see how many calories I've burned for the day, and determine whether I should be eating heavy, light, or eating nothing the rest for the rest of the day. You could probably make your own app much better than mine, tailored to your own data. Using local models to find trends that work for you can be beneficial. I'm trying to put as much data into my own app as possible to see if local AI can pick out trends, such as how I recovered faster eating certain foods or slept better on certain days.