AI-generated strategy simplification

I would like to suggest that, in the same way the AI chooses strategies using only one bet size, it could also decide on a simplification by organizing combos without any mixing in the strategy, losing as little EV as possible. This would greatly simplify the learning process.