Post

I remain impressed by @norvid-studies.bsky.social 's efficacy at prompting people to prompt their agents they've mastered prompt^2 engineering

@norvid-studies.bsky.social rfc

print 10,000 copies and start handing it out at major airports

anyone ever observed these image models are cracked?

how much material did you feed in here

this one is approximately the previous image + remix as gtd/7 habits with a dianetics type display and range of books then asking for a hyperproduced absurdist photorealistic vibe (picrel) and correcting back to cardboard you may also enjoy chat’s riffing, especially the personality audit

This is awesome, and for proper airport bookslop content there should probably be acrostic mnemonics, Claudes love those

I’m currently dual-wielding and only have imagegen through chatgpt but a wet opus pass on copy could work well

simcluster lore dot md when

I'm pretty curious how well AI written community cultural wikis are today, never really seen it tried but it seems: a) within the horizon and b) all the data is public. the 'hard' version would be to scrape everything and ask it to figure it out for itself, the easy one would be a sort of FAQ

funny you should ask

I think 'simcluster' is a bit of a lagging indicator as in most of the neologisms or "remember that thing when"s wont be tied closely to the word simcluster, you'd have to scrape a much wider amount of text. so maybe this is an argument for the slow burn FAQ "go find this" query format rather than

a one shot wiki. it would be interesting to see what it comes up with though.

idk how to serve this document but it got quite close to identifying lineage and current network

“idk how to serve this document”

refresh the post, incredible new era of workflow

it feels like the AI waterline is still thoroughly 'in the computer' in 2026 albeit rising daily but as it starts slopping around everywhere (social interaction, robots, new technologies) is when it's going to start feeling truly weird, imo