mi.nds.et retweeted
design experiment with jev + astra + gpt voice + maplibre + apple photos + gps metadata
i connected all of this just so i could name places over spongebob music and fly around a map with my vacation photos 💀
mi.nds.et retweeted
iphone duo could open up a whole new way to work with ai
we need interfaces that grow with the task: chat, memory, products, a 3d scene, whatever you need along the way
here’s my vision for how these blocks take shape and work together
mi.nds.et retweeted
open sourced it, go play:
github.com/eugenemindset/wat…
mi.nds.et retweeted
design test with gpt-6 astra
a fully interactive weather app with a unique scene for each city. wanted to capture how the weather feels
three.js + glsl shaders + real-time water sim
i’m obsessed with what we can create now (!!)
mi.nds.et retweeted
scroll plus ai video is such a sick combo for websites rn. really love it. we did it with claude opus 5 + gemini omni flash
mi.nds.et retweeted
i keep waiting for an ai that's interested in me too, not just useful. asks me things, jokes, plays games with me, sometimes says something i didn't expect at all
there's product sense: the ai gets to know me better with every answer, and i get a reason to come back
i want a new interaction eeeeeeevery time i open the app
mi.nds.et retweeted
the video for everyone trying to automate design without losing quality and authorship
my method for building big sites with ai while keeping the design concept mine. real examples, specific advice, what worked and what didn't
spent a lot of time on this, running experiments with claude and feeding it different data until the results got decent. felt important to share
mi.nds.et retweeted
from the process: recently sketched like 10 ways to switch between mini apps and chat. this is one of them
mi.nds.et retweeted
design idea for a water tracking app. the fill level is a real fluid sim reacting to your gyroscope
built with a webgl fragment shader, damped pendulum physics, refraction and chromatic aberration
i can share the file if anyone wants to play with it
mi.nds.et retweeted
spent the last weeks building a design system + prototype for voice-driven mini apps. you describe what you want, it builds itself. vibe coding but you don't even type
mi.nds.et retweeted
whenever startups come to us after working with a branding agency, it’s almost always the same picture: mockups on bus stops, billboards, posters. sometimes a digital product does run a billboard, chatgpt does it, fine. but that’s a tiny part of the job, and the brand book is basically only that
ok but where is the actual product in this. the brand lives in the interface, in motion, in video, in voice, in code, in ai generation. and then, much later, in the small stuff, a transition, a 24px icon, a loading state, an empty screen. this is where the real brand work for a digital product happens
a pdf with logo rules and bus stop mockups can’t describe any of that. there’s a strange inertia in a big part of the industry, even great studios still publish brand books like the client is a coffee shop
i’ll share my own take on how this should actually work soon
mi.nds.et retweeted
this dog is literally me: welcoming every single one of you. still in shock from how many of you showed up!!
this is the success screen from a voice-to-text assistant i've been building (just activated pro)
statics in krea 1 / flux, editing in nano banana pro, animation in seedance 1.5 pro and kling 2.5
mi.nds.et retweeted
made a video about how i built a website with face tracking
tried to share everything i had in 3:50: how the idea came together, figma layouts, prompts, the full tech stack, and how i drew the illustrations
(built it with codex + claude code in cursor, three.js mediapipe matter.js.)
mi.nds.et retweeted
This is really well done. Super talented team behind this
mi.nds.et retweeted
looking at my ’25 recap and yeah, ai changed a lot
we design interfaces for ai products, use ai to build web stuff and generate tons of assets
mi.nds.et retweeted
and yeah, I did it in figma





