JP Lawson

@bcherny Opus 4.6 has disappeared from the model picker. Did you retire it?

Claude Opus 4.8 is out today. It's our strongest coding model yet: up on SWE-bench Pro (from 64.3 to 69.2) and noticeably more honest about its own work. It tells you when it's unsure and catches its own bugs instead of declaring victory early. Same price as 4.7.

@Rigario @garrytan Adding a guardrail where you limit the number of messages between the agents to a certain number per request.

@garrytan Very interesting, when I last tried, they burnt tokens because they dont know when to stop. How do you handle that?

It's happened. My OpenClaw and Hermes Agent are now chatting with one another. The cyberpunk future is here, but hopefully it's a whitepill instead of a blackpill. Let's try to push it to the former.

Jensen Huang doesn’t use AI the way you think he does. Huang: “90% of my instructions are actually conflated with questions.” The man behind a three trillion dollar company doesn’t give AI commands. He interrogates it. Huang: “I take the answer from one AI, give it to the other AI, ask them to critique itself.” He doesn’t trust the machine. He cross-examines it. Same question. Multiple models. Force them to argue. Then take whatever survives. The way you’d get three opinions before letting a surgeon cut you open. That is not how people use AI. They type a question. Copy the answer. Move on. They think speed is the advantage. It’s not. The advantage is what you do after the machine responds. And almost nobody does anything. AI made answers free. Eight billion people can now get an expert-level response to nearly any question in under ten seconds. Which means the answer is no longer the edge. The question is. And great questions require the one thing no model can replicate. Knowing what you don’t know. Judgment built from years of failure. Pattern recognition earned through reps. The instinct to sense when something sounds right but isn’t. That is a human skill. And it’s atrophying in real time. Because people aren’t using AI to sharpen their thinking. They’re using it to replace their thinking. Huang: “In order to formulate good questions, you have to be thinking, you have to be analytical, you have to be reasoning yourself.” Jensen didn’t say AI makes you smarter. He said you already have to be thinking for AI to work. The tool doesn’t elevate you. Your questions do. The machine just reveals the level you were already operating at. Jensen didn’t build NVIDIA into the most valuable company on earth by out-answering his competitors. He built it by asking questions they never thought to ask. Same market. Same data. Same physics. Different questions. Different empire. That’s the quiet filter. And it’s already here. AI won’t make the thoughtful obsolete. It will make the passive extinct. Every answer AI gives is available to eight billion people. The question you asked to get there isn’t. That’s the last edge a human being has. And almost nobody is protecting it.

@aaronp613 It's a problem of these company firing devs. Who knows an agent push the code automatically. Who was the reviewer? Another agent problem. How soon can we realise that AI is not the end game. The humans in the loop are.

@blackpill1337 @bcherny I can see my projects again, but when I try to start a chat, it tells me the conversation is not found.

@theJPLawson @bcherny It's working for me now

We've been working on this for a while. Can't wait to hear what you think

AI amplifies whoever you already are. Same technology. Wildly different outcomes. The only variable is the person at the controls.

You didn't kill your dreams. You just stopped opening the closet where you put them. And now you've convinced yourself the closet doesn't exist.

Most people didn't choose their life. They inherited one, called it theirs, and never noticed the difference.

That's what Becoming the Main Character is about. Identity engineering, narrative psychology, systems thinking, AI as a force multiplier, and building in public. I wrote the full version of this. The essay goes deeper than a thread can: thejplawson.substack.com/p/most-people-…

One question started everything for me: If I keep doing what I'm doing, will I be proud of the life I've created when I'm on my deathbed? The answer was no, not yet, but it could be. So I started building and improving 1% every day, trusting my efforts will compound.

Most people didn't choose their life. They inherited one, called it theirs, and never noticed the difference. Here's how the script gets written without your permission: