Oliver Drobnik

Self-employed developer on Apple platforms, father of 2 girls, Tesla-driver. Side projects include AgentCorp (Swift-based Agent Framework)

[Oliver Drobnik @OliverDrobnik]

I’m consolidating my accounts! Follow me at @Cocoanetics for all future updates. This account will soon be deactivated.

[Gergely Orosz @GergelyOrosz]

From a (now former) Apple engineer:

“I really really wanted to work on something AI-related… it was just not possible to get a transfer to an AI-related team internally. So I went out on the job market. Got offers from two AI infra companies, accepted one. V excited to start!”

[Anthropic @AnthropicAI]

We’ve received notice that the Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5.

We'll begin restoring access tomorrow, and will share an update soon.

We’re grateful to our users for their patience, and to everyone who worked with us on redeploying the models.

[Andrej Karpathy @karpathy]

Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

[nic @nicdunz]

thank you chatgpt

this is very helpful

[Tibo @tibo_maker]

I turned the leaked Sam Altman and Mira Murati texts into a short film with Revid 👀

[Andrej Karpathy @karpathy]

Judging by my tl there is a growing gap in understanding of AI capability.

The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.

But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.

So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.

TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.

[Figure @Figure_robot]

Today we're showing Helix 02 that can tidy a living room fully autonomously

Figure is designed so when you leave the house, your home resets exactly how you like it

[Sam Altman @sama]

Peter Steinberger is joining OpenAI to drive the next generation of personal agents. He is a genius with a lot of amazing ideas about the future of very smart agents interacting with each other to do very useful things for people. We expect this will quickly become core to our product offerings.

OpenClaw will live in a foundation as an open source project that OpenAI will continue to support. The future is going to be extremely multi-agent and it's important to us to support open source as part of that.

[Andrej Karpathy @karpathy]

I took delivery of a beautiful new shiny HW4 Tesla Model X today, so I immediately took it out for an FSD test drive, a bit like I used to do almost daily for 5 years. Basically... I'm amazed - it drives really, really well, smooth, confident, noticeably better than what I'm used to on HW3 (my previous car) and eons ahead of the version I remember driving up highway 280 on my first day at Tesla ~9 years ago, where I had to intervene every time the road mildly curved or sloped. (note this is v13, my car hasn't been offered the latest v14 yet)

On the highway, I felt like a passenger in some super high tech Maglev train pod - the car is locked in the center of the lane while I'm looking out from Model X's higher vantage point and its panoramic front window, listening to the (incredible) sound system, or chatting with Grok. On city streets, the car casually handled a number of tricky scenarios that I remember losing sleep over just a few years ago. It negotiated incoming cars in tight lanes, it gracefully went around construction and temporarily in-lane stationary cars, it correctly timed tricky left turns with incoming traffic from both sides, it gracefully gave way to the car that went out of order in the 4-way stop sign, it found a way to squeeze into a bumper to bumper traffic to make its turn, it overtook the bus that was loading passengers but still stopped for the stop sign that was blocked by the bus, and at the end of the route it circled around a parking lot, found a spot and... parked. Basically a flawless drive.

For context, I'm used to going out for a brief test drive around the neighborhood to return with 20 clips of things that could be improved. It's new for me to do just that and exactly like I used to, but come back with nothing. Perfect drive, no notes. I expect there's still more work for the team in the long march of 9s, but it's just so cool to see that we're beyond finding issues on any individual ~1 hour drive around the neighborhood, you actually have to go to the fleet and mine them. Back then, I processed the incredible promise of vehicle autonomy at scale (in the fully scaleable, vision only, end-to-end Tesla way) only intellectually, but now it is possible to feel it intuitively too if you just go out for a drive. Wait, of course surround video stream at 60Hz processed by a fully dedicated "driving brain" neural net will work, and it will be so much better and safer than a human driver. Did anyone else think otherwise?

I also watched @aelluswamy 's new ICCV25 talk last week (https://t.co/RdaM23kvez) that hints at some of the recent under the hood technical components driving this progress. Sensor streams (videos, maps, kinematics, audio, ...) over long contexts (e.g. ~30 seconds) go into a big neural net, steering/acceleration comes out, optionally with visualization auxiliary data. This is the dream of the complete Software 1.0 -> Software 2.0 re-write that scales fully with data streaming from millions of cars in the fleet and the compute capacity of your chip, not some engineer's clever new DoubleParkedCarHandler C++ abstraction with undefined test-time characteristics of memory and runtime. There's a lot more hints in the video on where things are going with the emerging "robotics+AI at scale stack". World reconstructors, world simulators "dreaming" dynamics, RL, all of these components general, foundational, neural net based, how the car is really just one kind of robot... are people getting this yet?

Huge congrats to the team - you're building magic objects of the future, you rock! And I love my car <3.

[Oliver Drobnik @OliverDrobnik]

@manniefabian VV

[Oliver Drobnik @OliverDrobnik]

@manniefabian BBB

[bygen. @bygen_ai]

@sama it's fun lol

[Oliver Drobnik @OliverDrobnik]

@MatthewBerman @GroqInc Does Cursor support Agent mode on other LLMs?

[Matthew Berman @MatthewBerman]

"fix failing tests"

came back 20 min later after putting my son to sleep to find all my tests are passing

[Elon Musk @elonmusk]

Yes

[Jon Stokes @jon_stokes]

Confession: I thought "vibe coding" was a dumb X thing until I did it today with Claude Code and shipped an amazing refactor with it.

I've had my head in the problems with this code for over a week, so I could guide the model through the decision tree and really direct it and micromanage it in a way that actually felt like coding but on acid or something.

Anyway, $28.12 in tokens later, I have a 2,300 LOC PR that cleans up a ton of the app and is impossible to review without also using Claude Code to go through it and understand all the changes.

I just tried this method of code review, where I gave Claude Code a prompt like this:

"I'm on a branch and what I need is a complete code review for this branch. You'll need to diff it from `main`, examine all the changes, and then first give me an overview of all the changes in the code. Then ask if I have any questions or want to dive deeper on a particular change. Then you may have to walk me through specific changes. I also want to hear your opinion of the changes -- are they sensible, do they make the code better or worse and why, could they be improved, are there any coding standards in our app that they fail to meet, etc."

From there, it was basically like playing one of the Infocomm text-based adventures, where I just sort of explored around the PR, gathering even more insights and understanding even more of what was happening.

Anyway, TIL vibe coding is real, and if you're not doing it then you will lose to people who do. (Sorry, but these are the facts as of right now in 2025. If you're some sort of coding purist, you should plan to retire ASAP.)

[Matthew Berman @MatthewBerman]

SaaS is dead.

I can literally build any SaaS application in hours with AI.

Is it as polished as a VC-backed company with 1,000's of employees? No.

But it doesn't need to be. It's hyper-customized for exactly what I need. And it can evolve with my needs in real time.

I'm going through all of the SaaS apps I pay for right now, stack ranking them by complexity, and building them starting from the bottom.

Taste and data are all that matter now.

[Oliver Drobnik @OliverDrobnik]

@adcock_brett Can I borrow one to test :-)

[Spencer Schiff @SpencerKSchiff]

This was one of my early sketch-to-SVG tests. The gap isn't always this dramatic, but GPT-4.5 is often much better than its competitors at this kind of task. It's the only model that has come really close to this reference image (first try) and I gave the others several chances.