Paul Sanders

@zuess05 This idea you just uttered: what Saas allows you to push it to all comms channels at once? Challenges you, asks you follow up questions then publishes to 17 comms channels? None. Of course you can build it yeah I know... But will you? "Everything's already built" is an illusion

Silicon Valley predicts that AI will create the first "one-person billion-dollar company" very soon. But if Claude allows literally anyone to build complex software over the weekend for $20... Who exactly is going to buy their product to make them a billionaire?

@itsjoaki First of all.... Duh Second of all... @posthog

Google Analytics is one of the worst products a great company has ever made.

@jaydwivedi_ Minimalism. The death of culture and history

We lost another brand to minimalism

@thekitze will smith 🚮

suggest your top 3 mute words and i'll add them

@PaulSolt @kit Can't wait to dive in. I heard a lot of unhappy comments about kit's recent pricing changes. For modern/agent-first email, I'm also looking at @nitrosendx

My workflow with GPT-5.5 is completely different from GPT-5.4. I revived a 9-year-old app and pushed conversion from 0.5% to 7.1% in two weeks. Wrote up exactly what changed in this week's AI Code Brief. @kit super-easy-apps.kit.com/posts/how-i-bu…

@davis7 Okay... It's time for a "DavisBench - by Ben Davis"

This is one of many reasons I hate the "5.5" name so much. Every single 5 series model has needed at least medium, usually high reasoning to function because the "base" model was pretty bad. Most of it's power came from the insane reasoning stuff they baked in Should you use GPT-5.5 on low reasoning for everything? No. But it's pretty damn close, like 90% of the time, and I'm gonna keep pushing this hard b/c the natural tendency of everyone (self included) is to use the "best" tool always Always try low first, if it can't do it then bump from there We now have a good base model, so reasoning often becomes an anti-pattern as the model gets too much time to over think itself into verbose complex solutions when it knew how to just do the thing u wanted Again, this is not a ".1" bump, this is an entirely new foundation and I'd argue series of models. I know they can't call it GPT-6 because if that model doesn't end up being what mythos is hyped to be and more the entire fucking economy dies, but it's definitely the beginning of GPT-6, much like what 4.5 was last year just way better I think that's the reasoning behind the ".5" name, it's halfway to the new series which would be fine if we didn't just do 5, 5.1, 5.2, 5.3, 5.4 so it feels like another increment. It's not. Another thing is speed, this thing is fast as hell with low reasoning and feels amazing. Highly recommend trying it out with pi, feels so good

@quantumaidev @Rasmic

@Rasmic Was it written by Opus 4.7?

gpt 5.5 struggling to understand my codebase

@thsottiaux @ziwenxu_ This is Codex

@ziwenxu_ Didn’t know Codex was her

Sorry Claude, but I found someone new and her name is Codex. She's fully autonomous, smarter, and doesn't give me attitude

@hnshah Open-source first of all...

If Slack were built from the ground up today, what would it be like?

@shadcn 🐐

The whole point of shadcn/ui is I don't decide for you. New: when you create a new project, look for the switch to turn on cursor: pointer for buttons. Also available via the cli. npx shadcn init --pointer

@thsottiaux @TheRealAdamG I already have credits and I'm still able to work back and forth, even if subscription rate limits are exhausted, and I still see this. Somehow it doesn't detect that I have credits and I'm actively consuming them. As you said, small paper cut.

It’s the little things that matter, what are some small papercuts you have noticed in Codex? We’ll fix as many as possible in the next week.

@shadcn 🙋‍♂️ 🙋‍♂️ 🙋‍♂️ 🙋‍♂️ 🙋‍♂️ 🙋‍♂️ 🙋‍♂️ 🙋‍♂️ 🙋‍♂️ 🙋‍♂️ 🙋‍♂️ 🙋‍♂️ 🙋‍♂️ every single time. In globals.css. Any button, any role=button element including submit inputs.

I see. How many of you do this? I’d like to get a sense of how common this is.

@TheGeorgePu Claude Code

Vibe coding has been a thing for over a year. My friends built: - Fitness apps - Marketing tools - Casual games - Dashboards All shipped to TestFlight or GitHub. Now try this: name one vibe-coded app you actually use. Not tried. Actually use. I can't name one. Zero. They built. They kept adding features. They never figured out who it was for. We have more apps than ever. Fewer users per app than ever. Vibe coding didn't create a software boom. It created a software graveyard.

@EmerieOnoh If you can detect the pattern, you can prompt against it. :)

Okay… these ChatGPT designs are starting to look all the same.

@thekitze Especially those who would actually pay for creative work. They have neither an idea what to type nor the time to think about it. There's so much to build on top of a very capable model. Standards are gonna go up too.

billions of people faced with the textarea of the most powerful image model ever released, and have no idea what to type in there's a lesson there

@justalexoki They're not afraid Anthropic is gonna eat them and their search business... Now OpenAI on the other hand... Check the stack of Anthropic: entreprise and coding. OpenAI: everything: entreprise, media and creative, pro-sumer, ad-funded consumer, voice, realitime, workspace,...

wait why is google investing in anthropic isn't that a literal competitor

@BoyuanChen0 All I gotta say is good luck! It's an incredible model.

We are committed to continually improving the GPT Image 2 model! I am actively fixing various issues from the community feedback. Just reply or DM me your GPT conversation! Features like 2K or 4K images are already available via the experimental API. Hope you enjoy the model!

@paulg Well he's presenting alternative math, as it is his right under the first amendment.

This is who's running the government.

@thejeffcallahan Love the title and cover of the book!

@PaulSandersSCP Hi Paul!

Confidence-Maxxing Live 20 Zero introspection is good? x.com/i/broadcasts/1…

@picoito @emanueledpt x.com/blueemi99/stat…

@emanueledpt any skill for ui?

my codex 20$ sub about to expire anyone got any actual hands on benchmark feedback on gpt 5.5 vs opus 4.7? I'm trying to decide between 100$ codex or 100$ claude