@dillon_mulroy @badlogicgames we have a slack channel at work for this and itโs a gold mine
English
Andrew ๐ค
127 posts
Andrew ๐ค
@andamonium_dev
๐งโ๐ปSoftware Engineer ๐คAI hostile takeover waiting room
British Columbia, Canada ๊ฐ์
์ผ Mart 2023
403 ํ๋ก์24 ํ๋ก์
@badlogicgames this inspired me.
i now have a weekend project to start maintaining a journal for my clanker when it does stupid things
English
ladies and gents, i give to you: GPT 5.4, with 1M context window, context usage at 40%. brilliant shit. feeling the AGI.
English
Andrew ๐ค ๋ฆฌํธ์ํจ
maybe itโs good i grew facial hair and head hair. i looked like a hot dog that did contract killings in eastern europe.
English
Andrew ๐ค ๋ฆฌํธ์ํจ
Garry has a point
His point is completely wrong, but he still has it
Garry Tan@garrytan
Iโm trying to teach software engineers that their sacred cow of โLOC is badโ is probably wrong in agentic engineering You wonโt believe that in another 5 years
English
bread butter and cheese is just peak food. everything else is extra and unnecessary
English
Gave this a try and have now made it strongly-recommended for engineers at my work. It takes a mostly flushed out project plan and takes it that next step to being more maintainable code in the future.
Matt Pocock@mattpocockuk
My 'grill-me' skill went viral. mattpocock/skills is up to 9K stars. Quote tweets of it are doing numbers. It's the most useful skill I've written, and I use it even outside of coding:
English
@wesbos would love to see a walk through on how you set this up, been looking for a similar concept for live scoring
English
As someone who's mostly used nextjs for personal projects ive been experimenting more and more with tanstack router and now tanstack start. and holy moly why would anyone choose anything over this. It was so easy to set up SSR. The dx you get with vite as well. Incredible stuff
English
Disappointed by the response from the community to Lee's posts. He's been nothing but transparent in every post.
Lee Robinson@leerob
I'm a big believer in open source, especially as AI improves. It was a miss to not mention the Kimi base in our blog from the start. We'll fix that for the next model ๐ Their team clarified our usage was licensed in the tweet below. x.com/Kimi_Moonshot/โฆ
English
SWE-bench results are really cooked, you can now trivially see which models have been contaminated and will spit out exact answers.
The best way to measure model quality seems to be both offline benches + online evals. Think we have landed on something pretty good here!
Cursor@cursor_ai
We're sharing a new method for scoring models on agentic coding tasks. Here's how models in Cursor compare on intelligence and efficiency:
English
maybe if some of us had the option to toggle "max thinking" the world would be a slightly better and more expensive place
English
@Shreyassanthu77 @dillon_mulroy Yeah they were acquired by Cursor and now sunsetted. supermaven.com/blog/sunsettinโฆ
English
@dillon_mulroy Ohhhhh I did not know it was for existing users only damnn
English
i think iโm back to wanting a really good tab model - any progress here outside of cursor (i donโt have access to supermaven) and for nvim?
English
Was missing some clean-white images for a e-commerce store. I fed nano-banana an image of someone holding the hat in their hand and then a reference image of what I wanted and it pumped out exactly what I was looking for. Bananas.
English
With all the drama going on, it was incredibly refreshing hearing from an engineering leader and his take on ai, agents and focusing on the product.
Also, you don't need all the skills and mcp's, you can just build.
Wes Bos@wesbos
Today we have @southpolesteve giving us the scoop on how he made Vinext. His opencode setup and planing approach was super interesting
English
@asidorenko_ What if this isnt showing that you get the same accuracy with less thinking but rather the ceiling for accuracy is less then 60%
English
