AI-triaged reader feedback
Feedback on this site comes to me by email, never gets published, and never gives spammers a soapbox. But I read every note and often update a post because of one.
The idea: let an assistant summarize each incoming note, draft a possible response, and cluster related suggestions — so I can act on substance quickly and credit the person who raised it. Human stays in the loop; the AI just does the sorting.
Now working. A private admin page reads every note and, on demand, hands one to Claude for a summary, a sentiment/category/priority read, a suggested reply I can edit and send, and a credit line for when a note improves a post. It suggests; I approve. Next up: clustering related notes into a single “here’s what to act on” digest.
Have a correction or a different take?
I read every note and credit good ones in updates. It comes straight to me — nothing gets published automatically.
Send feedback