Prompt engineering is not engineering and it’s not hard, LLM ops is not ops if you are using OpenAI (which the question states you are using). If your solution has code for rag & prompts + a saas boiler plate, you have no moat. Whatever you have is already or will be eaten by openai itself (Altman keeps saying that as well).
(Still assuming you are using OpenAI) if your solution implementation has 2% prompts and AI specific stuff and 98% other functionality while AI is still central to your solution then you have possibly a moat. Seems existing systems adding AI are long term AI money makers.
(Still assuming you are using OpenAI) if your solution implementation has 2% prompts and AI specific stuff and 98% other functionality while AI is still central to your solution then you have possibly a moat. Seems existing systems adding AI are long term AI money makers.