This thread reminded me of something that I was thinking about, but have been too lazy to try yet
First, take something like https://www.nvidia.com/en-us/ai-on-rtx/chatrtx/ (or any other locally hostable LLM system, but apparently this is one of the better ones, if you have nvidia).
Create a dataset made up of the AGS manual, open source AGS projects, and solved questions on the forums (this last one perhaps being the most ethically grey, can skip if needed). Load it in, and...hey presto, you've got your own little AGS helper.
First, take something like https://www.nvidia.com/en-us/ai-on-rtx/chatrtx/ (or any other locally hostable LLM system, but apparently this is one of the better ones, if you have nvidia).
Create a dataset made up of the AGS manual, open source AGS projects, and solved questions on the forums (this last one perhaps being the most ethically grey, can skip if needed). Load it in, and...hey presto, you've got your own little AGS helper.