We’re using our general-purpose real-time observability tool to manage and tune our OpenAI implementations. I wrote a blog post with an overview, which might be helpful to those who want to know “I made the best prompt and put it on my server – now what?”
