when generating images with gemini/chatgpt ask for a grid of variations instead of one image, especially for simple illustrations you can vectorize in figma. better results and saves the hasstle of having to spin up a bunch of new chats. image models degrade fast after the first

v excited about this

in the pre-ai era, maintaining brand guidelines and design systems often wasn’t worth it for small teams with <2 designers. they went stale fast and the maintenance competed with shipping new work. now organizing them and making them accessible to ai agents is 1000% worth it and

this 👏. organized chaos is very a necessary step in going 0 → 1 on a new product idea. i’ve tried vibe coding without doing any upfront work with an artboard and it feels like a piece of my brain is missing and the product suffers

The ability to quickly reset and recover. From a bad interaction. From a bad day. From a missed workout. From a poor decision. You can start over whenever you want. You can't always control what happened, but you can control how long you carry it.

claude code chrome plugin + pinterest board is a killer way to iterate on a brand design skill

the hard part of AI coding isn't writing code. it's that most things you'd want to improve don't have a number. "are my docs good?" is not a metric. "is this test suite trustworthy?" is not a metric. "is this API reliable?" is not a metric. you have to build the ruler before

claude code is now generating ui directly in chat while we're riffing on a new app concept 🤯

wish you could disable chatgpts “pick between two responses” it breaks flow and the differences are almost never worth evaluating

x.com/i/article/2035…

love claude code's /btw feature - it's a great way to peak under the hood especially during lengthier tasks

would love shared context between claude.ai and claude code. i use the web interface to explore the shape of a project (especially great with inline ui now) then move to cc, but being able to go back and forth with full context would be amazing.

my uncle is a professional clown (he went to clown school). he once had an assignment where students had to embody the complete opposite of themselves from the moment they left their front door until they arrived at class. my uncle is warm, smiley, wickedly funny. so he put on a

experimenting with animated pixel art components - gemini for the sprites, claude for animation

way more bang for buck generating NxN grids with gemini/chatgpt image models vs trying to one-shot a single image