LLM layer for summarizing existing ML features #20604
andrewm4894
started this conversation in
Ideas
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
It would be feasible I think for each metric correlations call to get processed and summarized by an llm.
So in addition to the filters and ordered list, perhaps there could be some native way to also have an llm summarize the results and deliver that in the UI as part of the results in some natural feeling way.
Same for anomaly rate calls too perhaps for AR% and anomaly advisor.
Sort of like a post call hook that optionally helps summarize and make sense of the results form the traditional ML models.
Eg an llm might be better able to help augment stuff by for example leveraging it's internal statistical knowledge for the MC calls to layer on top of the results and similar for anomaly detection it might be better able to pick of all the AR elevated stuff what might be more important beyond just raw anomaly rate.
Maybe this could be a paid feature in that it could be layered in some way on top of the existing features in a more additive way. So like an optional add on.
I think at its core it's kinda mainly smart prompting and some clever UI/UX so assuming people ok with sending the data it might not be too hard.
Or maybe this is all just like an entry point to the native MCP stuff within netdata UI perhaps, eg start a chat from the results and have these prompts as part of netdata MCP maybe.
Beta Was this translation helpful? Give feedback.
All reactions