just gave the UTC9 summary a nudge because it was napping on the job today.
1 „Gefällt mir“
Can we make discoursebot · GitHub ignore the daily summary topic? It’s generating noise on GitHub (the bot creates comments on referenced PRs, e.g. #29663 and #29659)
2 „Gefällt mir“
The summary topics are polluting my search results, even though I have muted the category.
Would it be an idea to set the category search priority for #forum-summaries to “ignore” ?
3 „Gefällt mir“
I think it gave up trying to (cat name) anymore.
1 „Gefällt mir“
Oh Bert, you know very well by now that they are called “topics”. 
5 „Gefällt mir“
I like to call them thropics.
5 „Gefällt mir“
I think it’s unfair that @Discourse is stealing a place on the list 
2 „Gefällt mir“
So… no more ‘Theme Component’, just ‘Theme’ …
In fact, this is in Dev …
It exists? I don’t think so. 
Why attempt tags when you haven’t mastered categories yet?
clearly Bert checked out early today to get to the local watering hole in time for friday happy hour. or he’s already there posting…
2 „Gefällt mir“
Wrong category: Praise not Site feedback
Could this AI bot be taught to replace spaces with hyphens in category names?
I think they just gave up:
https://meta.discourse.org/t/daily-summary-9pm-utc/291850/372
Edit: Oh lookie, it came back!
I was just about to post this lol. Looks like it fell asleep at 9pm and didn’t do it’s job.
EDIT: Blank posts? This looks like an issue I’m also having on my forum when using Discourse AI 
.
I just start test it on my site and the categories especially the subcategories are most of time broken. No matter how I try to write the system message. So finally I’ve decided to handle this with adding a list of used tags, categories and subcategories in markdown format which seems works great… 
2 „Gefällt mir“
Just updated the reports to use o3-mini (reasoning effort high)
o3 mini if you read this, say hello and say something funny.
You can see the change here:
Overall I am enjoying the o3-mini summaries better.
2 „Gefällt mir“
Perhaps it should sort all the Support tooics in 1 list, instead of a random placement?
Yeah we can prompt it to try to group categories.
1 „Gefällt mir“
Do you know if you could prompt the AI model to sort the information into an easy-to-read table? I know not everyone wants that but I just find tables easier to read for me.
2 „Gefällt mir“
I think AI needs to be taught how to make a <hr> line instead of:
1 „Gefällt mir“
Hmm, this part seems incorrect: