pouria
125 posts
@levelsio @arvidkahl Reminds me of how Google ships 4 different products to solve the same problem
English
@arvidkahl I never liked it and always believes it should just be a regular post but then improve regular posts so you can do better formatting and put images etc anywhere
English
What happened to Twitter Articles? They did that $1mil for the Top Article contest, and now I rarely see any of those long-form reads here anymore.
English
thinking how to spend 50K in cloud tokens in a month, ideas?
English
@rileybrown if you want comparable performance (speed, model size, etc) to the APIs then you gotta spend 200k minimum. That's why I'm telling you to just use the APIs for now
English
Ok my new budget is 100k will make many YouTube videos on this. Who wants to sponsor?
Riley Brown@rileybrown
If I had a budget of $10,000 to build a local setup to run GLM 5.2 locally AND any future local model that comes out soon what should my computer set up be?
English
@marckohlbrugge @cursor_ai Cursor actually does sorta maintain one but there must be a reason that they don't release it in the app itself
github.com/cursor/cookboo…
English
It's surprising to me that every developer is making their own "kanban to manage agents" but none of the major IDEs (@cursor_ai etc) have this UI built-in
English
@rileybrown I looked at getting a setup for our company and it's honestly not worth it unless you're okay with the model being quantized. Just use APIs for a couple more years
English
If I had a budget of $10,000 to build a local setup to run GLM 5.2 locally AND any future local model that comes out soon what should my computer set up be?
English
@jackfriks it's really a shame because stuff like supabase and vercel had amazing dev experiences. Too bad dev experience just doesn't matter as much to my agents
English
now that AI is really good i feel its become silly not to use the complex solutions that are 100x cheaper in terms of infrastructure for software.
example: i avoided using cloudflare r2 before cause the docs for my other provider were easier to follow and implement... now i just tell AI: use this and make it work, it does it.
English
@jackfriks @tsvtxt can't tell if youre waking up at 4am or going to sleep at an ungodly hour
English
This man has been my inspiration for a long time.
Huge respect.
The lesson for all of us: Never give up!
jack friks@jackfriks
i just hit $37,000/month on my solo startup, so now i can call it a success i think but it got me wondering... how many times have i failed so far? and what's my success/hit rate? 📋 23 projects since 2019 🫙 18 misses ❌ 9 abject failures ✅ 5 HITS!!!
English
@theo Last time this happened in codex @thsottiaux named his first born after me
English
If this happened with Codex they would have refunded affected users and given everyone at least 2 resets
ClaudeDevs@ClaudeDevs
Earlier today, ~3% of Claude Code Max and Pro users hit a bug that showed an incorrect weekly usage limit, and in some cases blocked them from sending messages. This is fixed, and we're resetting 5-hour and weekly limits for everyone affected. Apologies for the disruption.
English
@jackfriks @yacineMTB Gdocs so I have the data there for when google’s AI is finally good
English
@yacineMTB but also apple notes goated for personal
g docs for business crap
English
@MatthewBerman For LLM as a judge, does it work a lot better if you have a council of LLMs as judges?
English
There are two kinds of loops:
> deterministic (ex: when site speed less than 50ms)
> LLM as judge (ex: when the website is fast enough)
LLMs are becoming fantastic judges. This opens up significantly more knowledge work to be looped.
Matthew Berman@MatthewBerman
Just launched Loop Library - a curated list of agent loops you can use right now. Find loops, submit your own, tokenmaxx!! signals.forwardfuture.ai/loop-library/
English
Devs, how much has your workflow changed in the last year?
curious to hear your thoughts on a few things:
- still using VSCode, or fully moved to Cursor / Claude Code?
-what’s your daily ratio of typing code vs prompting?
-what do you actually do while the AI generates?
Are you actually more productive?
English
people are just throwing around the word taste and expecting it to mean something
English
@pouriaa @LearnDomaining @arihant @tdinh_me My privacy policies are all separate pages on my single domain bigfatapps.com. Haven’t had an issue with app review and lets me maintain only a single domain instead of 8. Domains feel overrated these days for apps
English
@arvidkahl I’ve been thinking about writing one but I don’t wanna spend 6h on it just to get 11 views. May as well post to a blog and drive traffic that way
English
@evilissimo @Adidotdev I like GPT5.5 better but most of my team prefers Opus 4.8
English
@pouriaa @Adidotdev Isnt 5.5 already better than Opus 4.8?
Fable is different though. However didn’t get to use it myself bc it got scrapped before I could.
Now GLM 5.2 though WTH really good for an open weights model. Give that a spin.
I rather would go for GPT-6 🤣
English
GPT-5.6 and Fable 5 both drop tomorrow,
you can only use one.
What's your pick?
English
