Community-made working plans would be a killer feature.
But I do agree with your assessment. Each exercise needs a categorization (compound, isolation), compliments (if an exercise is a push, then what are some pulls), companions (if you're working arms at the cable stack, might as well do a bunch of arm/shoulder/back cable exercises), and a est. time to perform (including warmup, setup). This will allow plans to be generated in a way that makes sense.
Though, I think community made exercise plans are a better solution than trying to devise algorithms to generate good plans. Though, an LLM integration might work well for beginners, send a prompt with a list of exercises and goals (i.e., beginner looking for a 3 day a week strength plan, build one using these 20 exercises).
My next question would be why are we trying to use algorithms to generate good plans? Good and simple plans have been around for decades that are easy to find.
But I do agree with your assessment. Each exercise needs a categorization (compound, isolation), compliments (if an exercise is a push, then what are some pulls), companions (if you're working arms at the cable stack, might as well do a bunch of arm/shoulder/back cable exercises), and a est. time to perform (including warmup, setup). This will allow plans to be generated in a way that makes sense.
Though, I think community made exercise plans are a better solution than trying to devise algorithms to generate good plans. Though, an LLM integration might work well for beginners, send a prompt with a list of exercises and goals (i.e., beginner looking for a 3 day a week strength plan, build one using these 20 exercises).