End-to-End Tracing & Debugging [AIObs] NVIDIA NIM

[MichalBojko]

I am looking at the [AIObs] NVIDIA NIM dashboard and top 10 expensive prompts.
Some of the prompts are repeating.

Do we have or do we think about having deep dive analysis on the results?
For example, for repeating costly prompts, the product could provide:

• cost + ROI + several other topics analysis
•  result could be to suggest semantic caching implementation when affordability criteria are met