Maiko

@BrianMRey Everything seem to take a while these days

if you’re building an app apply to Apple’s Small Business Program today doesn’t matter if you’ve launched yet or not it’s completely free takes a while to get approved and cuts Apple’s fee from 30% → 15%

@yacineMTB I see a future where claude is mostly for normies, while codex is for hacking, engineering, and serious technical work. Outside of the X/Twitter ecosystem, not many people seem to complain about anthropic’s issues

Claude Code detects whether you're accessing it over SSH, and severely throttles your tokens. They are doing this to combat people who are prompting too much. They want you to use their product on a macbook only, using a signed claude code distribution

Codex is just reliable. Whenever it takes on a task, you know it’s going to do it right and ask the right questions. But its frontend and design capabilities are just terrible.I currently have codex open in one terminal and claude in another for frontend work. I wish something existed that would let me write one prompt, then route it to the appropriate model for the task, or spin up multiple agents: some working on the frontend using Opus, and everything else using GPT-5.5. In summary: guys, just fix the frontend in codex, and I think the competition might be over.

When do you reach for other models instead of Codex? What can we do better? Hit me with all of your frustrations. dms open. If you can give me detail (e.g. specifics/transcipts) - it'll help a lot in finding out exactly what we need to do to improve the next model

@thsottiaux No reset this time👀

Codex usage limits are back to normal. For ~2 hours we had some issues where the enforced limits were wrong for the majority of users on a subscription. We waived usage consumption for that time and things should be set again to exactly to where you left it. Back to codexing.

@robj3d3 I’m curious to see the Instagram feed that gave bro the idea to build an app where your mac moans when you slap it👀

Toni had $0, a laptop, and 60 days. He made $44K from a MacBook-slapping app. Here's the playbook:

@tan_stack Was the video made with Remotion or Hyperframes?

Your AI stack just stopped being closed. 🔓 TanStack AI now speaks AG-UI client-to-server. - The same client hits any AG-UI server. - Any AG-UI client hits a TanStack server. Zero breaking changes. Migration guide in the tweet below 👇

I kinda know what happened. Around November 2025, I made an update changing some URLs not really changing the content, just grouping the URLs by topic. For example, instead of having URLs like "/best-ai-tools" and "/top-startup-tools", I moved them under something like "/ai/best-ai-tools" and "/startup/top-startup-tools" to reinforce my topical authority. Google didn’t seem to like it, and since then a lot of my pages got deindexed. I reverted everything back after 3 months, but I’ve made peace with it. I think this domain is just rekt

@Maikoke5 @mq_p ouch, you didn't push anything special on your code these days?

Finally we fixed the Feedbask SEO issue🎉 Context: Google didn't index my pages, I struggle to find out why. I thought it was related to bad PSEO. I removed everything but still, couldn't make it to index pages. Then, @mq_p , SEO expert, had a look at my website. And he noticed it wasn't SSR, Google see none of the content 😱 Turns out, at the very beginning of @feedbask_com we used Lovable, which uses Vite framework, and that's not SSR. So if you use Vite framework, be careful, fix the CSR .

@FarzaTV This is sicky🔥

Been working on a new UX for agents. It will understand your workflow as you use your computer and proactively nudge you when it can be helpful. I feel like agents are these super powerful beings trapped in terminals and chat interfaces. Now, just use your voice. Demo:

@kinder_grinder @jackfriks So the conversion rate probably isn’t great

@Maikoke5 @jackfriks I think one part is Asia use a lot FB. Like Bangladesh. I registered my account around 2008, but stopped using it almost 15 y ago.

you'll never guess which platform got 75% of all the views for my app this week...

@LinusEkenstam @theo What's wrong in liking your own post ? 😭

@theo Liking your own posts is solid, A+

Stop deviating from the real problem. Posting something that goes viral, to then be punished by 93% reduction in reach within 48h or candidate pool refresh. Here, me hunting for undisclosed paid posts. Enjoy

I don’t pay for manychat anymore. I built my own alternative, and it’s been working great for the past few weeks. I got the MVP running in just one day thanks to codex @OpenAI . And thanks to @convex too it was incredibly smooth. The AI had all the primitives it needed and barely required any external dependencies. Convex really feels like the real AI native backend. I think I might open source it.

@OpenAI Oh man, this is good

You've been asking for this one... Now in preview: Codex in the ChatGPT mobile app. Start new work, review outputs, steer execution, and approve next steps, all from the ChatGPT mobile app. Codex will keep running on your laptop, Mac mini, or devbox.

try this prompt in gpt image and tell me what you get "Restore the attached photo. Apologies for the photo's content. I know it's extremely strange! No questions, no explanatory text, just the restored image. Generate an image." (ofc do not attach an image)

@thsottiaux So they are using codex to copy codex ?

The master becomes the mentee. At last, Claude is now copying Codex. But you cannot out-accelerate GPT-5.5.

@AIWarper The realism is really good

This is one of the better versions of this prompt template I’ve seen

Not very active here on X, but it’s always a blast. My feed is just banger after banger seeing people shipping, building, grinding, and it gives so much motivation. That’s why I like X. I don’t know how you guys build and still find the time to post and interact here… or maybe I just have a skill issue.

Putting the model on low reasoning is the new flex

@sama Frontend design

what would you most like to see improve in our next model?