i switched over to claude code desktop recently and started using cloud sessions. before doing anything else i would have a conversation with claude, have it create issues on GitHub (tickets). the reason for these tickets is for Claude to keep a history of work completed and work that is pending. at the same time, in my codebase i would have verification loops set up like unit tests, browser agents, typechecks, lint, etc. then have it start working on those tasks and open PRs with preview links so i can review the changes whenever itβs done. it will automatically fix any build errors or CI failures
vibe coded this feature section on prominent's landing page π
i feel like one trend i'm seeing with vibe coding is replacing images or graphics with actual ui
no more stitching together images on external design tools
just build the real thing with code and plop it right in π€·ββοΈ
a few years ago, i had this idea for desksetupcreator.
i wanted to build a 3d desk setup builder where you could design your dream setup using real products from real brands.
people loved the idea. it got around 3,000 sign ups to the waitlist pretty quickly.
but i made one big mistake.
i picked a project that was way beyond my skill set.
i didn't know three.js. i didn't know blender. i couldn't create 3d models. every new feature meant hiring someone, and it quickly became too expensive to keep going.
eventually, i had to put the project on hold.
fast forward to today.
instead of trying to build the perfect product, i rebuilt the idea around something i could ship. an ai-powered desk setup image generator.
it's still early, but it's already brought in 230 visitors and 35 sign ups over the past month.
no paid conversions yet, but that's okay. i'm mostly excited that people are finding it again.
sometimes the best version of an idea isn't the first one. it's the one you can actually ship.
hot take:
i don't think local ai is popping the ai bubble.
yes, local models are getting better.
but after experimenting with them, i realized running truly capable models for coding, image generation, video generation, and other heavy workloads requires an insane amount of compute.
the average person simply isn't going to have that kind of hardware sitting at home.
if there's an ai bubble, i think it's more likely to come from companies struggling to turn massive ai infrastructure costs into profitable businesses.
ai isn't going away. the question is who can build sustainable businesses around it.
seo is finally paying off.
cramberry just hit 442 visitors in a single day, the highest it's ever been. π
traffic has been climbing steadily, and we're at 7,530 visitors over the past month.
MRR dipped from around $500 β $400, but i'm not too worried. it's summer break, so conversions have naturally slowed down.
i'll take growing traffic every day. the conversions will follow.
i've been working on something new π
as more people turn to ai instead of traditional search, i think every business will eventually ask the same questions:
"is ai even recommending our business?"
"why are our competitors showing up instead of us?"
"what can we do to improve our ai visibility?"
that's what inspired prominent.
it's built to help businesses understand, monitor, and improve their visibility across ai platforms.
lots more to build, but i can't wait to share what's coming π