Max Weisel

Normcore 3 is out!! Our team has been doing incredible things with this release. I'm so proud of everyone who worked on this and all of the developers shipping titles with Normcore. We've hosted over one billion multiplayer games, and we're on our way to 1 billion more!

@mertdumenci try this: github.com/glittercowboy/… - spent an hour having it ask me every possible design question, then it spent the day and one-shotted an entire project

This has proven incredible so far. Opus is great for chatting with. 5.3 is incredibly thorough and trustworthy when you give it a detailed task. I much prefer CC+Opus when I’m debating or planning something, exploring a codebase - and then 5.3 takes it and builds it for me.

Today's meta: get Opus 4.6 w/ agent teams to write you a detailed plan. Interact with it, brainstorm. Then go to Codex 5.3 xhigh and say "implement it". 3h and going.

VCs earn more with a $1B fund that returns 2x (poor) than a $200m fund that does a 5x (exceptional). That is the root of everything that is wrong with VC today.

@infinitehumanai @bcherny run it in a VM with --dangerously-skip-permissions. this has been a massive performance gain for me. Write up an extremely detailed plan and then let Claude Code one-shot it without prompting for any commands.

@bcherny Yeah I've been through this but still hitting issues. Appreciate the effort on resolving.🤞

I'm Boris and I created Claude Code. I wanted to quickly share a few tips for using Claude Code, sourced directly from the Claude Code team. The way the team uses Claude is different than how I use it. Remember: there is no one right way to use Claude Code -- everyones' setup is different. You should experiment to see what works for you!

A few random notes from claude coding quite a bit last few weeks. Coding workflow. Given the latest lift in LLM coding capability, like many others I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in December. i.e. I really am mostly programming in English now, a bit sheepishly telling the LLM what code to write... in words. It hurts the ego a bit but the power to operate over software in large "code actions" is just too net useful, especially once you adapt to it, configure it, learn to use it, and wrap your head around what it can and cannot do. This is easily the biggest change to my basic coding workflow in ~2 decades of programming and it happened over the course of a few weeks. I'd expect something similar to be happening to well into double digit percent of engineers out there, while the awareness of it in the general population feels well into low single digit percent. IDEs/agent swarms/fallability. Both the "no need for IDE anymore" hype and the "agent swarm" hype is imo too much for right now. The models definitely still make mistakes and if you have any code you actually care about I would watch them like a hawk, in a nice large IDE on the side. The mistakes have changed a lot - they are not simple syntax errors anymore, they are subtle conceptual errors that a slightly sloppy, hasty junior dev might do. The most common category is that the models make wrong assumptions on your behalf and just run along with them without checking. They also don't manage their confusion, they don't seek clarifications, they don't surface inconsistencies, they don't present tradeoffs, they don't push back when they should, and they are still a little too sycophantic. Things get better in plan mode, but there is some need for a lightweight inline plan mode. They also really like to overcomplicate code and APIs, they bloat abstractions, they don't clean up dead code after themselves, etc. They will implement an inefficient, bloated, brittle construction over 1000 lines of code and it's up to you to be like "umm couldn't you just do this instead?" and they will be like "of course!" and immediately cut it down to 100 lines. They still sometimes change/remove comments and code they don't like or don't sufficiently understand as side effects, even if it is orthogonal to the task at hand. All of this happens despite a few simple attempts to fix it via instructions in CLAUDE . md. Despite all these issues, it is still a net huge improvement and it's very difficult to imagine going back to manual coding. TLDR everyone has their developing flow, my current is a small few CC sessions on the left in ghostty windows/tabs and an IDE on the right for viewing the code + manual edits. Tenacity. It's so interesting to watch an agent relentlessly work at something. They never get tired, they never get demoralized, they just keep going and trying things where a person would have given up long ago to fight another day. It's a "feel the AGI" moment to watch it struggle with something for a long time just to come out victorious 30 minutes later. You realize that stamina is a core bottleneck to work and that with LLMs in hand it has been dramatically increased. Speedups. It's not clear how to measure the "speedup" of LLM assistance. Certainly I feel net way faster at what I was going to do, but the main effect is that I do a lot more than I was going to do because 1) I can code up all kinds of things that just wouldn't have been worth coding before and 2) I can approach code that I couldn't work on before because of knowledge/skill issue. So certainly it's speedup, but it's possibly a lot more an expansion. Leverage. LLMs are exceptionally good at looping until they meet specific goals and this is where most of the "feel the AGI" magic is to be found. Don't tell it what to do, give it success criteria and watch it go. Get it to write tests first and then pass them. Put it in the loop with a browser MCP. Write the naive algorithm that is very likely correct first, then ask it to optimize it while preserving correctness. Change your approach from imperative to declarative to get the agents looping longer and gain leverage. Fun. I didn't anticipate that with agents programming feels *more* fun because a lot of the fill in the blanks drudgery is removed and what remains is the creative part. I also feel less blocked/stuck (which is not fun) and I experience a lot more courage because there's almost always a way to work hand in hand with it to make some positive progress. I have seen the opposite sentiment from other people too; LLM coding will split up engineers based on those who primarily liked coding and those who primarily liked building. Atrophy. I've already noticed that I am slowly starting to atrophy my ability to write code manually. Generation (writing code) and discrimination (reading code) are different capabilities in the brain. Largely due to all the little mostly syntactic details involved in programming, you can review code just fine even if you struggle to write it. Slopacolypse. I am bracing for 2026 as the year of the slopacolypse across all of github, substack, arxiv, X/instagram, and generally all digital media. We're also going to see a lot more AI hype productivity theater (is that even possible?), on the side of actual, real improvements. Questions. A few of the questions on my mind: - What happens to the "10X engineer" - the ratio of productivity between the mean and the max engineer? It's quite possible that this grows *a lot*. - Armed with LLMs, do generalists increasingly outperform specialists? LLMs are a lot better at fill in the blanks (the micro) than grand strategy (the macro). - What does LLM coding feel like in the future? Is it like playing StarCraft? Playing Factorio? Playing music? - How much of society is bottlenecked by digital knowledge work? TLDR Where does this leave us? LLM agent capabilities (Claude & Codex especially) have crossed some kind of threshold of coherence around December 2025 and caused a phase shift in software engineering and closely related. The intelligence part suddenly feels quite a bit ahead of all the rest of it - integrations (tools, knowledge), the necessity for new organizational workflows, processes, diffusion more generally. 2026 is going to be a high energy year as the industry metabolizes the new capability.

Used claude code and gamekit-cli to one-shot a minecraft game and then the `/add-multiplayer` command to make it multiplayer. These tools are getting insane. github.com/gamekit-agent/… #claudecode #gamedev #unity

@Azadux really? I've used the powershell one for editing the registry and making scripts to manage our CI. I haven't hit any restrictions. WSL one works great, but WSL 2 file system access can be slow if you're accessing files outside of the WSL filesystem.

@maxweisel I think the power shell version has a ton of user prompts due to sandboxing which makes it pretty inconvenient. Is WSL is to use? Haven’t tried it

Windows users… what’s your vibe coding setup like? Neither Claude code or codex cli support windows natively

Guys I think I figured out how to get the Harry Potter VR game back on track

With everything going on at Meta, Steam Frame might be the most important VR headset launch in a long time. Fingers crossed the content lineup on the device matches the hopium.

@waslegit but so once the AI hype train slows down, I wonder how much we're all going to have to pay to use frontier models

@maxweisel That sounds right, at their scale it’s probably more distributed but it has to be a massive expense. Amazons $8 billion investment should be able to cover some of the AWS costs lol

Does anyone know how much a $100/mo Claude subscription costs Anthropic to run? I’m convinced it’s $1,000+ and that at some point these tools are going to get unbelievably expensive.

aaaannnd upgraded to Max 20x. Problemo solvelemo!

😭😭😭

This is wild, after hitting the limits of the Max plan, I burn roughly $10/hr of tokens using Claude Code. At this point, it's going to be more economical to start buying additional Max plans.

Max'd out the Max plan. What am I supposed to do now? Go to bed?

@JordanKutzer @ThrowbackVR nice work!!

100K on YouTube for @ThrowbackVR!

@waslegit I imagine it would cost $100k+ for me to get the same hardware to run a single instance of Opus 4.5 and it wouldn't have enough throughput to run Claude Code for a team of 5 people, even if their queries were perfectly spaced out to not overlap.

@waslegit Based on the price of the GPUs they're using and the amount of throughput they get on this model, I feel like they have to be operating at a catastrophic loss.

also I’m talking about raw hardware cost here, not Claude API cost

@AaronCederberg This assumes that they charge you what it costs them to run the models? I feel like they’re operating at a loss?

Easy enough to test with API, no? All you gotta do is use the API directly with your own back end for a month and see how much it costs (I'd think API isn't subsidized like the flat rate plans are) Personally I'd be very afraid to do this. Pretty sure I'm costing them thousands.

@waslegit This assumes that they charge you what it costs them to run the models? I feel like they’re operating at a loss?

@maxweisel I use ccusage to see token costs if they were standard API requests. I haven’t run it recently but the most I’ve done is $500 on a $100 max plan, definitely a deal. github.com/ryoppippi/ccus…