Evolving optimized builds and build paths for off meta playstyles

The idea is you create a custom game with two equal skill bots (Ideally at your skill level) one goes a randomly generated build and another the "meta" build for that champion. You can limit the pool of items the random one can choose from depending on the type of build you want to get (Full Ad, Full Ap). Then you evolve a build or build path using Evolution algorithms that I won't go into detail about. In theory the outcome should be the best build/ build order in a one V one situation. Why? This would be meant for determining the best off meta build, as an AD bard main I want to know if my build is the best AD bard build and I think that this would be a very effective way to do so. The main issue i see with implementing this is that I can't specify what the bot buys and in what order.

I'm sharing this because I can and someone may find this interesting and because I really want to be able to control bots and this is an interesting application of bots. I would love to hear any related Ideas or questions and what you would do if you could make bots build certain Items.