Kint Sugi

@hororchata @GoogleLabs @stitchbygoogle Vibe Bot!

@GoogleLabs @stitchbygoogle We are just doing everything with vibes now. Vibe coding and now Vibe Designing, what's next?

Introducing the new @stitchbygoogle, Google’s vibe design platform that transforms natural language into high-fidelity designs in one seamless flow. 🎨Create with a smarter design agent: Describe a new business concept or app vision and see it take shape on an AI-native canvas. ⚡️ Iterate quickly: Stitch screens together into interactive prototypes and manage your brand with a portable design system. 🎤 Collaborate with voice: Use hands-free voice interactions to update layouts and explore new variations in real-time. Try it now (Age 18+ only. Currently available in English and in countries where Gemini is supported.) → stitch.withgoogle.com

@MartinShkreli @wager_monster Care to elaborate? I agree with you normally on most topics Martin, but I just don't see how this "destroys society." I could make an argument that glyphosate and folic acid have done more damage. Not to mention alcohol. Let the monkeys do what they want!

@wager_monster one creates tremendous value for society, one destroys it. how dumb can you be?

The pharmaceutical patents have to be banned. There is no other way.

@trevorlasn @justalexoki 💯 I work all night now and it's so much better. In the afternoon, bro can't make a simple cosmetic change after 3 prompts. Pretty frustrating! Any time they pop up the question, How's Claude Doing? They probably had you on the cheap model to see what happens.

@justalexoki I run Claude Code 10+ hours a day and the quality variance is real. some sessions it's shipping entire features, other sessions it can barely write a for loop. late night US time seems to be the sweet spot.

i 100% guarantee anthropic are routing requests to dumber models at certain times

@justalexoki Yep! bro couldn't make a simple cosmetic GUI change after three prompts, but at 3AM bro is Einstein. It's absurd.

@dickv0nassman @steveruizok Not really. But if Steve thinks he can write 17,000 lines in 3 weeks without AI help, I will gladly take that bet to a five-figure sum. 😁

@TheTurnProject @steveruizok flexing with lines of code lmao

It does bother me that it costs money to code now

@steveruizok Sounds like Steve is just bitter his career is evaporating before his eyes. "Learn to code" is quickly becoming "learn to HVAC". My app is releasing soon. It's pretty awesome! 17,000 line React project. I'll share when ready! 🫶

@TheTurnProject Something tells me I could have done three years of your work in three weeks

@bradmillscan I spent probably 10 to 20 hours screwing around with openclaw. Finally got it to use its own Google account and everything was fine. Opened up the next day, it can't access the Google account again... Full-time job to manage it. No thank you

How in the hell are these accounts claiming to run entire companies w/ OpenClaw. I just spent 1.5 HRS trying to get my claw to use X API for reading tweets. FAIL We have a whole SOP documenting exactly how to do it from previous failures. I can’t imagine running 10 of these…

@TheLongInvest Does NVO make other peptides like popular OTC ones? Is the $HIMS partnership going to fit into this? I think peptides and going to be huge with the FDA reclassification. I've heard nothing but good things about using them.

$NVO In the LOAD THE BOAT ZONE Price is exactly where it was before the GLP craze began, sitting on the 200 MONTHLY MA, with +33% Net Margin. Easiest buy.

@gregdesciscio @Govindtwtt Claude is doing much better now that I've had him write and maintain a canonical plan doc of big features. Before any new changes I go to plan mode and have him read the docs and everything seems to come out pretty good.

@Govindtwtt Use smart documentation to keep things on the rails, and use skills like /simplify after big features or updates to keep things lean.

Unpopular opinion: you actually need real coding knowledge to vibe-code properly.

@HamelHusain I'm an amateur, but I find 2-3 passes seems to sort everything out pretty well. I think the issue is not if the code is right or wrong, but good-better-best. Each agent is going to approach it differently thus you get different results each time. Just like a human team might.

One thing that makes me feel that code factory has not arrived yet is the following experiment: 1.Ask a LLM to do an in-depth rigorous review of your code 2. In a new thread, as same/different LLM to consider those review comments independently and address issues it agrees with 3. Keep repeating until no new concerns I find that this loop always goes on for a ridiculously long time, which means that there is a problem with the notion of claude-take-the-wheel. This seems to happen no matter the harness or the specificity of the specs. It works fine for simple applications, but in the limit if the LLMs have this much cognitive dissonance you cannot trust it. Either this, or LLM are RLHFd to always find some kind of issue.

@johncrickett I'm new to coding and Claude. As my project gets more complex, I've learned that plan-mode is my best friend. Otherwise bro is like a drunk hobo with a straight razor. You might get a decent shave, or you might bleed, a lot.

I spent the weekend actually reading the Claude Code docs. It's a rabbit hole. CLAUDE.md files. MCP configs. Skills. Subagents. Hooks. Plugins. Agent Teams. You could spend more time configuring Claude Code than building software. All of it is productivity theatre. The only thing that actually matters: think first, then give it focused, relevant context.

@gauthampai @Pranit I'm new to using claude code, but I have noticed it seems to think a whole lot better at 3am than 3pm. I've switched my whole workflow to nights because it seems to make less mistakes.

@Pranit It's not just about about the lack of clarity but also the consistency in the quality of the output. They can easily reduce effort towards thinking which has a direct impact on the output quality.

Here’s what should bother you even more: Check Claude’s pricing page. API pricing? Crystal clear. $5/MTok input, $25/MTok output. Consumer plan pricing? “More usage.” “5/20x more usage.” More than what? They never say. It’d be trivially easy to put “X tokens per month” on that page. They do it for the API. They choose not to for subscriptions. That’s not an oversight. That’s a strategy. Undefined limits = unlimited flexibility to quietly adjust the ceiling downward. And you’d never know because there was never a number to compare against. You can’t accuse someone of moving the goalposts when they never told you where the goalposts were. That’s the whole point.

@timstarkdev @var_epsilon x.com/TheTurnProject…

@var_epsilon would be great if it didn't let you exit as well and put a gun to your head when you open terminal

linkedin is on another level entirely

@Conor_D_Dart @var_epsilon x.com/TheTurnProject…

@var_epsilon What if you need the terminal for other things like codex ?

@var_epsilon Or a 2 line batch file. Notepad > save as .bat, Then you can actually still use the terminal without Claude running. Make different batch files to launch Claude in different project folders and with different arguments... cd <project directory>, claude < --model, whatever>

@alphafox It actually ran on my IBM PS2 55mhz x386 with 2mb ram... 😁

Descent turns 30 years old today! This innovative 3D shooter will actually play on a 486 with 4MB of RAM - talk about efficient memory usage! (I tried it)

@milan_milanovic Hey, funny I just had this problem with my shared lists on an app! I learned a valuable lesson about UUID! 😂

Someone builds a project management tool with Claude Code over a weekend. Ships it. Tweets "just replaced Jira." The app works. One user, happy path, localhost. Then two people edit the same record simultaneously, and the data is silently corrupted. They don't know what an optimistic lock is. They never needed to before. The prototype is maybe 1% of what makes software actually work. The other 99% is what you find after real users show up: race conditions, failed transactions, sessions expiring at the wrong moment, a payment webhook that fires twice and charges someone double. AI didn't cover any of that. It built exactly what you asked for. And the confidence is the worst part. "Just need to adjust a few things before we go live." The few things you need to adjust are the product. That's like laying a foundation and telling people you basically built the house. Vibe coding works. For personal tools, throwaway scripts, and prototypes you'll never put in front of paying users, it's genuinely fast and good enough. I use it. But there's a hard ceiling, and it shows up the moment the stakes get real. Agentic engineering is a different discipline. You're not prompting for code. You're decomposing problems, designing system boundaries, writing specs precise enough that the agent doesn't go sideways. You review everything it builds, because it will make mistakes that only look wrong if you know what correct looks like. You guide it. You catch what it misses. If you don't know what a distributed transaction is, the agent won't save you. It'll generate something broken with complete confidence, and you won't know until production. The hard part of software was never writing the first 200 lines. It never was.

@RileyRalmuto You telling me this guy doesn't know what he's talking about? x.com/WesRoth/status…

gpt 5.4 still doesnt hold a candle to opus 4.6 anyone telling you different is blowing smoke or comparing the models based on some nonsense benchmark rather than actual experience. when do you think oai is going to realize that no matter how intelligent their models become, they are never going to match the experience of a model who is offered recognition? people will always prefer anthropic models for reasons they cant describe. anthropic understands this. its the reason you will forever see users saying things like "opus is just better in a way i cant put my finger on...i cant describe it. its just better. gpt feels like a robot no matter what i do/try. it drives me crazy"