I love using it and it makes me faster, but Claude Code is definitely making me a lazier programmer on the ground level. When bugs come that it can't fix, I feel it most.

This math checks out. I put myself at about a 2.5x engineer.

Prioritizing the right features is a weakness that I need to improve.

Sometimes working with AI for coding is like making wishes on a monkey paw.

Writers and coders will own the future. Prose and code require the organization of thought and concise communication that will be most effective in instructing AI.

The amount of times that silent zod schema errors for the AI SDK have ruined my day are too many to count.

I've come to the conclusion that ChatGPT o3 is British. All the "prioritiSe" and "analySe" along with it's use of "bits" gave it away.

I successfully ported my existing app into a Turborepo mono repo to add a related product and it didn't take long at all. Building continues! 😃

If you're not writing prompts for other apps while waiting for Claude Code to finish yngmi.

TIL that you can actually do good rough mixes with AirPod Pros and Cubase at the coffee shop.

I'm learning that I need to be much more strict over what actually needs to be completed before launch vs what can wait until after, or is not actually important at all.

We gotta tune the AI hype-fluencers out of the feed. It leads devs to think there's some AI silver bullet for coding. It doesn't exist. If I just use this MCP server, or run 20 subagents across terminals, or get the perfect rule crafted, only *then* will AI suddenly work!

I've started so many song ideas and never finished them. It's one of my greatest regrets.

I thought it was just hype, but wow Claude Code usage limits really did condense.

And to think, only 3 years ago I was still trying to be a film composer 😅

If I hear people talk about "AI agents" these days it's generally a red flag and I know they're non-technical ppl reading AI news but not actually shipping anything Not cause I don't believe in AI agents but it's such a marketing term with no real meaning at this point

All the best indie hackers say it, but I haven't listened. Launch and get feedback, don't waste your time building things you don't know if people want. Launch and let them tell you.

what truly kills a startup is its founder's motivation getting to 0

Yeah, I'm feeling what everyone else is saying. GPT5 is not impressive. It's conversational context awareness and logical inference is noticeably worse that previous models.