First post here. hi. Found you guys through the vibe coding article.
okay so i did something possibly dumb three weeks ago.
I was spending $20/month on ChatGPT. using it for everything. Client emails, brainstorming, sometimes just talking through stuff when my brain felt stuck. Then i saw some guy on youtube running llama on his desk and not paying openai anything. Watched like four more videos. Ordered a mac mini M4 with 24GB before the last video even finished, I dunno what got me...
nine hundred ninety nine dollars. because youtube told me to.
(my wife had questions)
Anyway it arrived. installed Ollama which took maybe four minutes? pulled some model called Qwen because thats what eveyrone seemed to be running. and then i just... sat there. Watching it generate text. on my desk. no internet required.
it works?
Like actually works. not as good as Claude for the tricky stuff. i still use Claude for client work, anythng that needs to be actually good. but for quick drafts and summaries and explaining code? the local thing handles it.
But then something shifted.
i started typing things into it that i NEVER would have put into chatgpt. Personal stuff. anxious thoughts i normally keep in my head. notes from my therapist. questions about whether im a good husband. the kind of things you dont want sitting on some server in san francisco being fed into a training run.
its not smarter than Claude. but i talk to it more honestly.
is that weird? that the privacy part mattered more than i thought it would?
I keep seeing people building complicated setups. vector databases. RAG pipelines. home automation. Meanwhile im just asking Qwen why my chest feels tight somtimes.
am i using this wrong or is simple okay
genuinely asking. i have no idea what im doing but its working (i think?) and i wanted to check if other people ended up here too or if im the only one who bought expesnive hardware just to journal at it.
Anyway, thanks.
(delete if it's basic for this forum)
