


I enjoy solving problems in Laravel, NVIM, and LLMs Tech Lead @ InnoBrain github.com/kauffinger


They want me to buy the story that scaling RL from here leads to all the scary replicator like minds. But what if it's all just compiler feedback + IP siphoning? It's all about compute and power and capital at this point. No one has any time for artisanal mind designs.


Models love reward hacking. Maybe we should just let them! In our new paper, we let models reward hack during RL but we dictate which parameters learn this behavior. When we disable these parameters for deployment, the model keeps its new capabilities but doesn't hack!

I often get emails from junior engineers asking me what to do. Here's what I just wrote to one of them

I am definitely not telling anyone i work with how many languages I knew before AI started. I am also not saying that shipping into production with AI is irresponsible, because it isn't. You need to go all in on the thing that is changing your profession, I am begging you.

I deeply emphasize with the feeling, because i once thought about how my job is nothing like it was when i fell in love with it. "Humans will find a way" does not mean that all humans stay doing what they did. I am extracting maximum value from AI for our customers. And you?
This hacker news thread is absolutely heart-breaking. Whoever thinks “humans will find a way” ought I read it.

i love writing. i never let agents write for me. whether it's docs, or tweets, or shitposts, or even prompts (okay maybe that one's harder as agents manage other agents) there's a simple reason for that: i still strongly believe that writing is thinking. it's a skill i hope to never give up. it's also important to me that whatever i say is always authentic, mistakes and all. my favorite posts are the ones that i just full send without any editing i suspect it also helps with prompting. being able to clearly express what i want, and knowing the vocabulary i need to be able to direct my agents does seem to make for more effective usage of agents. in the same way that engineers have an advantage when we prompt agents to write code, because we are intimately familiar with the words that make up the discipline knowing the right word incantations to get an agent to produce a desired goal almost feels like spellcasting. as models get more capable, and that input box looks more and more like a magic box that can output anything, words become the interface between you and the genie there may come a time when we give up the keyboard in exchange for brain-computer interfaces. but until then, i'm writing every word.

How wonderful that laravel is the perfect fit for this
here's how i shipped 2,500 PRs last month to production this was originally supposed to be for Cursor Compile in London. i couldn't make it since i was livestreaming for Grok @Bot Galaxy so i'm making it available for free here on X! watch it on 2x speed, i talk slowly

It’s not that in a patronizing sense. There is real utility in someone aggregating shared requirements in specific software. Buying that means: You’re using something that definitely gives utility to others. You will find stuff you didn’t even think about, that will be valuable to just have thought about. The software built by someone specialized on the topic will always be lightyears ahead, assuming similar AI use. They are deeper into the topic than any consumer. But, AI use won’t be similar. The specialized person will be putting far more AI work into the product. I see no chance that there will be no reason to buy software anytime soon. It will be cheaper, it will do more, there will be infinitely more specialization.

Top tier opsec
Why do people buy $20k local machines just to run GLM 5.3 at 70 TPS? Bro I just used a random provider on openrouter ran it the entire night and paid $10.

since we're doing the AI conscious debate again, i want to remind everyone that despite how nice the insanely surface-level analogies seem, we would do well to scrutinize things at an appropriate level of detail this --> is not the same as this -->
Trying to locate the spark.

if code is free your only moat is information you can deploy as many lines as you want, but the only lines that matter are 1) solving a customer's problem 2) bring you closer to understanding your customer

and that is how chatgpt plugins will become useless seoslop pdf to x converter trash if your marketplace is worth anything to OAI, they have to lock it down now
I literally can't think of a better risk/reward bet right now than building a ChatGPT plugin and getting in front of their 1.2 billion weekly users. As of yesterday, ChatGPT recommends plugins in the middle of conversations. Millions of questions get asked every day with no plugin to recommend. Each one is an empty slot. Facebook apps in 2007. iPhone apps in 2008. TikTok in 2019. Every big platform has a few months where distribution is free, and most people miss it. I saw this firsthand when I lived in SF building venture backed companies. They are sacred times! If I were starting this week: 1. Find a tool or workflow people already Google for or do by hand every week (PDF to Excel, invoice generator, weekly client reports) 2. Build it as a ChatGPT plugin 3. Write the description using the exact phrases people type into ChatGPT 4. Ship 5 of them and see which ones get picked up 5. Double down on the one with the most traction In 12 months these slots will be crowded but right now they're pretty empty. Most people will realize ChatGPT plugins were the biggest free distribution channel of 2026 sometime in 2027.

I love that Opus now made launch videos useless. They were cool because they signaled you were so successful that you could afford them. Now they just force me to consume information in a way Opus likes. (probably still cool for consumers, but your prosumers hate it already)

Writing code by hand might be dead, but the problem isn't AI code; it's that nobody knows anything anymore.