and the training costs are going to EXPLODE (WSJ report)
does this mean it'll soon be done and they all need to switch to consumer? do I get this correctly?
B2C is a tough sell for AI, everyone wants it for free
how did OpenAI drop the ball like this, they had such a huge lead
also, dude over 50% of US businesses have already adopted AI, there isn't much growth left?!
just feels good to be kind to the AI sometimes
haha exactly
haha right, we forgot
😂😂😂
this is how the website of the future will look like
y'all probably don't have the correct jacket
if you're website doesn't have RTX mode
then what are you even doing 🤌
mindrudan.com
I am beginning to trust the Mythos benchmarks
no wonder they say it's a leap vs Opus 4.6
here's the actual level in April 💀
I suspect that fixing enough of these ultimately makes a difference between a good product and a sloppy one.
They add up and people eventually start thinking the product is unstable.
Very hard to change that mindset once it's there.
fixed a bunch of papercuts that I
a) wouldn't know about (runtime bugs)
b) were too small to put serious time into before
there's just something about Sentry's AI ergonomics that feels better than anything else.
it's always so damn spot on with the root cause and implementation, I am suspecting they have access to some unreleased SOTA model sometimes =)
✋ stop using commands in Claude Code
just use skills.
you can get the same DX with:
---
disable-model-invocation: true
user-invocable: true
---
this way you benefit from the entire skill ecosystem and also better cross-compact across CLIs, IDEs.
seems like commands are gonna be deprecated btw
always respond with 1 so Anthropic doesn't nerf models thanks =)
seems more relevant than ever =)
if you give your AI agent full permissions, at least do this
npm i --g trash-cli
alias rm='trash'
running my own Ralph sh is still my fav way of coding with AI, esp for exploration
highly recommend
and that is to be a government
OpenAI vs Claude vs Cursor uptime
there is only one correct choice really
ffs
Anthropic’s Internal DevOps Prompt Leaked
> carefully analyze our API availability and fix all the issues one by one. Run tests and Make no mistakes. Output no em-dashes.
<⏳Thinking...
Sam Altman confirmed internally:
👉 the definition for AGI never changed
it was always
"Anything Generating Income"
you can opt out of GitHub training here
github.com/settings/co...
literally the next page I opened
guys don't give your data, it's all we have left lol
🚨 if you've missed this
all your code (including private repositories) will be available for free in the next version of ChatGPT 💀
thanks Microsoft!
If a reviewer works a standard 8-hour day, 250 days/year, that’s roughly 51,000 working years
basically an entire career for 680 people working simultaneously from age 20 to age 95.
yeah I plan to prompt until 95 get with it
30 BILLION LoC added by Claude in 2026.
I skimmed through the code and yeah:
LGTM 👍
lmao Microsoft rejected me from some AI event
should have used Copilot more 💀
😆
