These are all fair points I agree with. Let's just say I might've started writing code without thinking too deeply about the problem...
About the last point (slowness), this is definitely a problem right now. In my case, also retrieving the information from Amazon takes a sizeable amount of time; for the LLM part I think it can be optimized with finetuned models and performance will generally get better in the future. So maybe overall slowness is the easiest problem to solve.
About the last point (slowness), this is definitely a problem right now. In my case, also retrieving the information from Amazon takes a sizeable amount of time; for the LLM part I think it can be optimized with finetuned models and performance will generally get better in the future. So maybe overall slowness is the easiest problem to solve.