
rege retweeted
Hire an AI teammate that does the work.
One agent across every channel.
β
Replies in seconds in the customerβs language
β
Tracks orders, books deliveries, and processes returns
β
Uses visual understanding to resolve issues
β
Remembers every conversation and improves weekly
Start free β yourgpt.ai
userinterface wiki β ui/ux rules across 11 categories
npx skills add raphaelsalaja/userinterface-wiki
impeccable β shape, critique, polish
npx skills add pbakaus/impeccable
emil kowalski β ui polish + animation taste
npx skills add emilkowalski/skills
ui skills β fixes spacing, hierarchy, motion perf
npx skills add ibelick/ui-skills
ui ux pro max β palettes, font pairs, ux rules
npx skills add nextlevelbuilder/ui-ux-pro-max-skill
shadcn/ui (official)
npx skills add shadcn/ui
frontend design (anthropic)
npx skills add anthropics/skills
transitions.dev β 44+ ui transitions
npx skills add Jakubantalik/transitions.dev
gsap (official, 8 skills)
npx skills add greensock/gsap-skills
no ai slop β makes ai writing sound human
npx skills add petergyang/no-ai-slop
9
369
agent mode in the mockup motion tool is insanely good π₯ wait for small glimps at the end π .
i just dropped a @YourGPTAI app screen recording in. it:
- picked the right interactions and scenes
- added focus, zooms, and smooth easing
- cut at the right times
2-3 tweaks. done in 30 mins
mockup motion tool free inside coanimator app - coanimator.com
it all comes down to this
- the model will get cheaper and better every few months
- your edge is workflow, data, and distribution, not the api call
- if a new model release kills your product, it wasn't a product
build products that get better with new models, not killed by it.
1
6
141
anthropic and openai both hit the same wall
git at agent scale π€
anthropic (scion): merges went from ~2.5k to 30k+ a day. history is ~100gb and ci runs ~25m checkouts a day. they replaced git, and checkouts went from minutes to seconds
openai (pando): kept git and rebuilt the backend. ~80% faster pushes, ~1pb of git data served a day, built in ~10 weeks with codex
We built Pando, a new Git backend for our monorepo, to handle more pushes and fetches without changing how our engineers and agents use Git.
Pando cut push latency by ~80% and now serves nearly a petabyte of Git data a day. We can grow Git capacity alongside agent usage.
7
473

