It’s time to JEPA pill the world!
awesome-jepa: A curated list of papers, models, code, datasets, and learning resources for Joint Embedding Predictive Architectures (JEPA), the self-supervised approach to world models proposed by Yann LeCun.
I created an auto-grill skill inspired by @mattpocockuk's grill-me
it lets an AI review its own work, find weaknesses, and improve the result before handing it back
would i trust it with life changing decisions?
no.
would i trust it to clean up code, refactor, improve copy, find edge cases, or polish a side project?
oh yes.
i just used it to refactor all the metadata across one of my projects using SEO best practices, it caught bunch of inconsistencies
sometimes AI workflows are just about eliminating the boring parts
this is your monthly reminder to stop arguing on X and start building a fucking product or service
it took years to get here
there are no shortcuts
lock in
Design is full of codewords. Knowing them changes what you can ask for, and what you can get back, whether you're working with devs, or an AI.
“tint this neutral color”, “fix this widow”, “nudge it to the optical center”
I wrote them down: https://t.co/aFyd5avj9o
I poured my 10 years of teaching experience into a skill.
It's called /teach, and it can teach you anything.
Here's how it taught me to solve a Rubik's cube:
"... every company is about to get the ability to hire infinite employees."
This is why everyone is getting serious about managing token costs.
Managers are accustomed to headcount budgets - but they're now realizing they can spend, sometimes without limit, on agent labor budgets.
This is obv not sustainable for either their company, or for the app providing the agent to them (if they pay a fixed monthly costs vs usage).
If you use poke check out my autmation to keep track of codex launch’s.
Maybe you’ll see @interaction mentioned in one of these one day.
https://t.co/a8F2tH8RUy
Today we’re introducing Builder, a new MagicPath plan for people working with Codex, Cursor, Claude Code.
For $10/month, get unlimited external-agent calls and a multiplayer canvas with visual editing, design systems, live links, Figma export, and more.
Introducing FrontierCode: a coding eval that raises the bar for difficulty & quality. Each task took 40+ hrs of work by leading open-source maintainers.
Models write sloppy code that works but isn’t maintainable. Our eval is first to measure: would you actually merge this code?