AI Developer

Maxed out 144FPS with 1000 Bots on the map, GG @threejs is INSANE!

@bcherny @joaogsantosc @AnthropicAI I would love one too :D Dedicated user in Cursor April 2025 to Claude Code March 2025 to today (most of my histories deleted).

@joaogsantosc @AnthropicAI 🫶

Chegou presente da @AnthropicAI / Claude Code Team 👀

Anthropic, despite having lost many of my chat histories since March 2025, still topping the token board here! WEN Package @bcherny 🤔

@StartupBros Nice! That said, I literally can't even submit mine xD! github.com/sculptdotfun/v…

more Claude code usage than almost anybody in the world and I still don’t have a swag box 😭

@WarforgeXP @Roblox @KayLousberg We are gearing up to release our first Roblox game, 100% AI coded also 🫡

@WarforgeXP @Roblox @KayLousberg Really nice work :)

My 15 year old daughter and I spent the last 7 days building this. She wanted a @Roblox game where you are a bad guy and build a dungeon. 100% of the code written with Claude Code Opus 4.8 + Roblox MCP. Many assets built with Roblox Cube. Other's purchased from @KayLousberg

@AzFlin Balance, UI, Juice, Fun...all pretty serious things most forget. 90% of the coding was done in our current game , but have now spent weeks on the above.

to all the people vibecoding games FYI the "coding" part of a game is very easy now and takes maybe 25~% tops of the work the bulk of it is artwork, assets + animations, map design, game design, and iteration and none of that is easily AI-able

@modestasraz And then u could watch split/overlay/1 after the other etc, def some cool stuff u can do

@modestasraz Not sure if watching the demos randomly in split would help (ok it would be decent), but if you did data analysis to find the tendencies and then could show the similar rounds/plays/set pieces, that would help for sure.

Game changer or not? I have a crazy idea in mind, imagine the top down view of up to 4 demos all in one view so you can watch like 4 demos at once. I imagine this would let you see player tendencies faster than rewatching demos one by one. #CS2

@morganlinton It's far better than 3%, benchmarks are not linear, nor are they always fully representative....Sure I can still get heaps done with Opus 4.8 and codex 5.5, but Fable was on another level!

While I do think Fable is a very good model. I think people are overreacting to the impact of losing access to it. Are we really suffering because a model that is 3% better than GPT 5.5, is unavailable for what will likely be one week? 👀

@maybe_im_sam @cursor_ai Nice work! AI + Perf = Win

@cursor_ai debug mode continues to be such a secret weapon. Just had it resolve a thorny issue where I was getting what felt like random little render lag spikes in my game. Normal perf sweeps and automated benchmarks that I have weren't catching the issue and it was hard for me to pinpoint what the exact cause was because the timing and situations to me seemed variable in each play test. The hitches would happen several minutes into gameplay which made it hard for automated benchmarks to come across it. Debug mode systematically instruments code paths and then asks you to try and repro. i.e. just play the game as normal and tell it when you saw the issue again. It then uses the debug logging to pinpoint causes (and rule out unrelated hypotheses) and then apply fixes. Once fixed it automatically will clean up all the instrumentation. I've found I usually reach for it specifically in scenarios where the root cause in't obvious or very hard to repro. e.g. the worst kinds of bugs...

@doodlestein Yep, just like the expert coders telling me things I am already doing with AI are impossible for the last 2+ years xD

The video in question (don’t be this guy!): x.com/apostructura/s…

It’s revealing to see so many medical imaging experts being dismissive towards Midjourney (although of course they all preface their dismissive comments by saying that they aren’t doing that… but no one is fooled), calling them naive, and giving all the reasons why it won’t work and won’t change anything. It sort of helps to explain why imaging is so insanely overpriced and rare in elective, prophylactic, discretionary contexts. Some of the arguments are so absurd. “Oh, water was used in the 50s, we walked away from that. That’s going backwards.” Really? They had sub-millimeter resolution back then, too? Oh. So maybe that’s totally irrelevant. It can’t see through bones? So take images from all around the bone. Sure, you won’t be able to see inside the bones, but you don’t need to for it to still be very useful. The criticisms all seem to make a common analytical mistake that I used to see all the time in the hedge fund business, which is ignoring key aspects of a disruptive change when applying an argument. If the cost of the scan is 99.9% lower, and it takes a minute instead of an hour, and there is no risk from radiation, then you can realistically do LOTS of scans much more frequently. That means you have a whole new modality to work with: time. You can look at changes over time. Duh. That’s a really big difference! This one aspect of the innovation dramatically changes the calculus, and suddenly most of the “fatal flaws” identified by the experts melt away into obvious irrelevance. Also, some of the flaws they identify have obvious next steps that could address them, like adding additional modalities (like light) to help differentiate tissue types. Anyway, it’s all a bit maddening to see this tripe, and also seeing all the people gleefully reposting and echoing it as if it’s some kind of “gotcha.” In 5 years, it will all seem as laughable as the video of that European space executive dismissing SpaceX and reusable rockets as a pipe dream 8 years ago. One of the great things about the free market system is that smart people like @DavidSHolz can just do this stuff and the old guard can’t stop them. Under a Soviet central-planning system, David would be dismissed as an ignorant dilettante with dangerous, naive, wrong ideas. He would be laughed out of the room, kind of like how the Russians laughed at Musk when he tried to buy some of their rockets before deciding to build his own.

Picking one agent is not smart...I ran an analysis of my code base before launching our next game soon, and Claude Code and Codex both found issues each other missed. Swiss Cheese Method wins!

@levelsio Now imagine Fable 5 game jam...

You can see the FULL ranking of ALL 945 games on vibej.am now The ranking gets more muddy as you go passed the first 100 though so don't take too much from it, but it's nice to see!

@levelsio And refactor/review often

Good point and yes partly it's that and people vibe coding But you can also just ask AI to add more test so things keep working well You can vibe code all you want as long as you test after

@0pusdev @CharlesUchi_Dev Looks exactly the same, are you saying you are re-creating it 1:1?

@CharlesUchi_Dev I first create their world, then I make buildings for myself. My buildings are my character. Another world, based on messenger logic.

x.com/rapkuryers/sta… Remember! It all started like this! My test day is over, and I've already started drawing graphics similar to those in this game. I told everyone I could do it, yes, I have a lot of work to do, but I'll do it! Look at what I've already accomplished.

@ARB1TERzzzzz Insane, is that CS_Office?

搬磚搬久了就會畫磚頭了😭 MECCHA CHAMELEON

@modestasraz That would be a really nice achievement :)!

Before and after for the CS2 demo viewer smokes. Not in production yet. Current goal: Figure out how to replicate the CS2 smoke filling areas.

@threejs Or 3 months....or with Fable 5 with a few weeks 🤔

What if the AI is able to do this in 3 years?

Maybe add a way to auto detect things like games and auto hide, auto hide when recording (like with OBS), warning if no output in mic detected at start after X seconds (I have recorded for a while, but my Mic died through NVIDIA Broadcast). Faster way to check my history of transcripts, or search, or export (I was able to retrieve it using AI, it's on my PC, but it's very slow to scroll back). Sometimes the history even has huge gaps in it, like it might have todays, then the next is like 1 week ago.

Calling all haters of @WisprFlow - give me your biggest issue with Wispr. Yes I will personally read through each and every comment and have our team right some wrongs.

@mattshumer_ I just use my Agent Workspace with Tail Scale github.com/web3dev1337/ag…

Is there a good way to control Claude Code on your computer (with dangerous permissions) from your phone, like Codex Mobile? And no, I'm not using Dispatch, it sucks.