Jan-Peter Franke

@TukiFromKL This benchmark is only half of the story. But yeah still impressive

🚨 Let me make sure you understand what DoorDash just did.. They built a $70 billion company by paying people $3 to deliver food in the rain.. Now they're paying those same people to film themselves doing chores.. mopping floors.. folding laundry.. loading dishwashers.. Not for content.. for AI training data.. they're paying you to teach the robot how to be you.. so they don't have to pay you anymore.. First they paid drivers to prove people would order food to their door.. then they started testing drones and robots to replace the drivers.. Now they're paying people to record their own bodies doing physical tasks so the AI can learn the movements... this is the Uber playbook on steroids.. > Use the human to build the demand.. > Use the human to train the replacement.. > Remove the human.. DoorDash are paying you to film your own obsolescence.. and you're going to do it.. because rent is due..

@ivanburazin Yes, very true. I was talking to my colleagues about that as well. I definitely see it coming.

New category emerging: Headless SaaS Not infrastructure as a service / platform as a service Traditional software (Photoshop, Slack, Jira) rebuilt with agent-first APIs. - No UI - Programmatic access - Essentially the same product with different interface Entirely new business model.

@karpathy This looks sick! Would love to experiment with that hardware 😅

Thank you Jensen and NVIDIA! She’s a real beauty! I was told I’d be getting a secret gift, with a hint that it requires 20 amps. (So I knew it had to be good). She’ll make for a beautiful, spacious home for my Dobby the House Elf claw, among lots of other tinkering, thank you!!

@TheGlobalMinima This is very true. The reason why I have my own custom system that no openclaw, perplexity computer or manus could match. Custom to me and with core features that no other system has.

the harness. the harness. the harness. This is what anyone in the agentic space talks about now. The harness is what turns the science of llms to the engineering of it. The model is no longer the moat.

Been rocking on 5.4 the last few days but have to admit that I had to re-tweak it a lot to get something good out of it without a huge amount of bullet points. After doing that and a few other things, it's almost there to match opus... almost. The first days felt like my agent's character „died“ 😅 Weird how much you're affected by the „character“ of your agent... for coding it‘s kinda different, there I’m fine with codex or Claude code out of the box.

Use GPT-5.4 mini and nano for most of your Openclaw use cases. It’s a workhorse model.

@shazcodes Okay calm down 😂

I rebuilt twitter’s entire backend in 45 seconds using claude code. no, I don’t know python. No, I don’t know what an api is. if you spent the last 10 years learning syntax, I’m genuinely sorry for your loss

@MatthewBerman Same... Going around and my head is just like: ahhh interesting can I build that with ai?

I’m genuinely struggling to think about anything but AI. My mind constantly goes back to it. It’s becoming a problem.

@svpino Well mine doesn't act like that ,😅

Claude is retarded. All of these models are. I wanted to schedule a skill every day at 8:00 am. Claude decided to schedule it at 7:57 am "to avoid the on-the-dot" surge. I SAID 8:00 AM! Do the darn thing the way I asked you to do it! You gotta be crazy to trust these models.

@rezoundous Better get roasted than ignored! When getting roasted, you can roast back or at least learn from it.

What hurts more for a founder? 1. Getting roasted 2. Getting ignored

@steipete Good idea. Get rid of the junk and enjoy the clean feed.

My openclaw twitter mention block cron job is working unreasonably well. Turns out AI is really good at detecting spam/reply guy/promo stuff. Runs every 5 min and cleans up my mentions - I actually see useful replies now and Twitter got pleasant again!

@kreative_gabe @marcelkargul lol true

@marcelkargul It looks like the exact design as the one in the ragebait video. x.com/kanikabk/statu…

this wasn’t a prompt. this was: - 8 hours of work - 5 iterations - 3 senior designers

@droidbuilds Not really… only if you can’t imagine something unique

every product already exists. every tool already exists. so what do we even build now?

Depends if your idea is already solid and if there are mockups. I would say UI first in a lot of cases. At least to a certain point. Reason is: it's easy to change things around until you're happy. When it's already full stack that gets slower. And backend first is difficult unless you have a good plan of what you really need.

When building an app, what do you make first? The UI or the backend logic?

@toddsaunders Logitech MX Keys and a Stream Deck are good. But starting to talk to the model might be a good idea as well.

Okay now that I’m in the “agentic engineering” world. I need a new keyboard. Mines slowing me down. All my claude pilled friends, what are your recommendations?

@marcelkargul lol that design is not even outstanding. Very doable if you know what you do. If I look at some of the software that is actually released and maintained by humans, AI would have been a 1000x better choice in UI UX terms. 😂

people say AI will replace designers. please, show me an AI that can design like this.

@sickdotdev Damn, I always knew that OpenAI 4.6 thing would be better than me…

You idiots think you can code better than OpenAI 4.6? Bro, the US military literally used it to capture Maduro. They wrote a prompt and one shot that dude and you really think it can't code your fucking shitty productivity app. You're actually high level retards. Put your ego aside and wake the fuck up. It's better coding than you will ever be.

@T_Zahil Definitely

I'm sorry for all the developers who will hate me, but I F*CKING LOVE "CODING" WITH AI

@qrimeCapital Be creative and find new ways. That’s actually the way it is nowadays. You come up with stuff first, and the time delta until it’s done elsewhere is the room of potential.

Anthropic just one shotted my 200k ARR business today. I have reason to believe Anthropic saw the agent skills I had been making that create interactive charts based on RAG model strategy and user defined learning material. I had hundreds of customers and half of them cancelled their membership today. I made this learning textbook with interactive simulations and charts last night for my girlfriend who is studying for an interview. Not sure what I’m going to do now.

Devs acting like they didn’t write slop code before AI. 🤡

@aryanlabde True story

Vibe coding is so addictive. You open your laptop “just to fix this one bug” at 5pm and suddenly it’s 2:30am.