How much should an AI fitness product remember about the user?
I’ve been thinking a lot about the difference between personalization and real adaptation in fitness software.
A profile with age, goal and experience level is useful, but it still treats each workout request almost like a new session.
A more interesting model is longitudinal: remember what the person actually completed, which movements and muscle groups were trained recently, how difficult the exercises felt, what restrictions exist, and how nutrition has looked over the previous days.
Then the next recommendation is based on that history instead of starting from zero.
The hard part is finding the right balance. Too little memory makes the product repetitive and generic. Too much poorly structured data can create noise and false precision.
For people building AI or health/fitness products: which user signals have actually made personalization meaningfully better in your experience, and which turned out to be unnecessary?
Replies
I’d split memory into three layers: an immutable event log (completed workout, load, rest time), a derived state that decays (fatigue/readiness), and explicit user constraints that never decay without consent (injury, equipment, accessibility). Then require every remembered field to answer: which next decision can this change, and for how long? That makes retention testable instead of defaulting to ‘store everything.’ Sensitive nutrition or health data should also be visible, editable, and deletable by the user.
The completed-versus-self-reported distinction upthread is the right one to build around. Difficulty ratings drift the moment someone works out what they're used for, same as most self-reported signals once they become a lever instead of just information. I'd treat anything safety-related, an injury or a hard restriction, as a rule the system should never quietly override, and leave the rest as memory that's allowed to update. Building that split in from day one is a lot easier than retrofitting it later.