I just realized...

you are not meant to have breathing space

"oh ya let me finish this project first"
- you're ngmi

"I'm so busy"
- you're ngmi

I don't think we need another LLM, what we need is a better stackoverflow

- you ask a question
- ai agent tries to answer
- then other people can view/comment or give a better solution than the ai. (Peeps with a superiority complex are gonna love this lol. Just display "questions solved ai couldn't" publicly on profile page)
- then you can flag it as "ai solved" "ai failed" or "human solved"

Stackoverflow was like a cool community of new devs and seniors but they ruined it with "duplicate question" + other moderation crap. Watch primeagen vid

This is what it will look like
- dev1: asks a nextjs related question
- ai: gives correct answer
- dev2: comes in and lectures dev1 on why he should switch to svelte

Trying to solve other people's problems or just glancing at them every now and then was a great way to improve coding skills.

The ultimate coding skill is debugging.

Better Auth featured on Forbes - https://www.forbes.com/sites/dariashunina/2025/06/09/the-most-promising-startups-from-the-first-ever-yc-spring-batch/

Note to self: Suffer more

https://www.dontbuildthis.com

Literally provides no value whatsoever

This is like a form of procrastination from actually building and shipping, there's no technicality no formula for this shit

I instantly wanted to build an alternative to this: ongjustshipit.com but that also defeats the purpose, you shouldn't be talking to llms or anyone for that matter you only need to ship it as fast as possible and solve your own problems

I might do this

registration deadline is today if you're interested

prize is 5K

It's Sunday

I'm not sure what to ship

I finished the backend for my main project which is better-wakatime, now only the frontend left but I don't want to rush it

And while I was designing the frontend penpot kept crashing so now I'm making a desktop version of it with tauri

btw if you just want to convert a website to a desktop app there's a great tool for this called Pake. It uses tauri under the hood and creates <5mb desktop apps

This was not a pleasant experience

I couldn't build the frontend, kept getting JVM errors. Which could have been avoided had they provided a shell.nix file

Then thought I could just self host the whole thing then found out that the docker image for the backend is 500mb. no thank you. its basically exaclidraw with extra features, make this make sense.

Building for web first is a mistake!

but seriously wtf? why should my experience be like this. both as a potential contributor and as a user

why are companies making products web first. make a great native product then you can make the shit web version an option.

If telegram was web first, none of us would be using it. Native first, then web.

So I'm throwing in the towel, not worth it. I will be sticking to pen + paper + imagination for designing from now on. I'm serious. My one great take away from this is that I don't need software to design software.

Was literally about to post "power went out didn't get to build anything 😭" and it came back