rishav

my primary constraint is now code review and architectural understanding. designing systems by-hand helps me internalize the intricacies of the thing i've built. i can map my intents to specific code. i can (overenthusiastically) tell you why my solution is better than yours. i usually prefer hand-rolling architecture/design patterns and *then* letting LLMs materialize it in code. design is the fun part for me. recently though i've been impressed by system design work from agents. it's often better, more nuanced, and explores paths i wouldn't have considered within the same timeframe, but i'll never feel one with it like i do my own work. i always review LLM code/implementation and try to conceptualize patterns, becoming one with the model and all, but my mental model is less durable than something i built myself. is it worth sacrificing this granular understanding in-favor of being able to verify the correct outcome instead? i think eventually economic pressures with lean the industry here, but how do software/eng folks think about this today?

take me back to the 1.7.2 days

what if it's called mythos so claude subs getting access would remain unreasonable claude = for plebs mythos = for nonplebs strategic positioning in the name of safety

@OfficialLoganK @Starlink honestly apartments + new constructions should build in starlink on roof by default functioning internet is basic infrastructure like water or gas lines

I moved today, tried to setup internet, provider needed to send a tech, earliest appointment was Wednesday. Got a @Starlink Standard 4 delivered and online in less than 1 hour, truly sci-fi, ty to SpaceX team and Elon.

@theo this is certainly plausible by post-training the model at various reasoning/response lengths based on value in the prompt unclear what the intent is, as there's many simpler options to control output lengths, but it's technically viable

Fine I'll take the L here. Just repro'd this 3 times. I should stop assuming competence from Anthropic.

Fun fact: LLMs have zero idea how they are configured. They don't know what GPUs they're running on. They don't know what temperature or reasoning level they have set. They don't know if they've been quantized or not. They're just doing next-token prediction. As always.

@hnshah it's langchain right

Without naming names, some of the AI-powered tools I try to use feel like you need a PHD to use them.

basically, but to capture your "essence" you'd have to document many small details about your qualia, thoughts, interactions, actions. easy for written or spoken ideas, but how to capture complex internal ruminations in obsidian? i only ask because it's surely possible (in various levels of fidelity) but can't require extra conscious effort to be practical. for example neuralink requires notable baseline mapping, and even then captures low signal patterns not complex thoughts. how much diverse information derived from yourself will models need to generalize *you* holistically? imagine a turning test but for yourself? am i talking to myself or an artificial version of myself?

Yes it's the tractable form of brain upload. There's a ton of scifi on brain uploads that requires way too exotic tech (scanning and simulating brains etc), when we're about to get a lossy and approximate version of that *a lot* sooner via LLM simulators. You can easily imagine a "brain upload" startup - you show up for a few days to carry out detailed video interviews, then they use all that data with an LLM finetuning process to "upload" you and give you an API endpoint of your simulation that you can talk to. Look at what's already possible with HeyGen as an example, but combine it with an LLM model that has deep knowledge and personality. Trippy and admittedly kind of dystopian but in principle quite possible around now.

As I build my own 2nd brain 🧠 on Obsidian using @karpathy ‘s wiki idea, it suddenly dawned on me - one day when we r gone, our kids could inherit an interactive map to your mind, passion, obsessions, work, fascinations… It’s kind of beautiful way to think abt your 2nd 🧠.

@alxfazio simply don't run tests, easiest to find out the hard way

@NousResearch @garrytan open source lives on though

@NousResearch @garrytan ayo tried explaining this, got canned by the nous girl

The most underrated thing in AI agents right now is: OpenClaw/Hermes Agent is just more free than other locked down AI agents (the standard out-of-box Claude/ChatGPT route) "Free the Claw" is not a vibe I understood until I tried it. Now that I have it, I don't want to go back.

@DrizzleORM understandable, enjoy your day

Drizzle v1 is too dangerous to release

@thekitze @superdoteng +1 innovation per tab nowadays asu people know what's up

@superdoteng wow very unique ui, finally some innovation tbh

Meet Superconductor. A native macOS app for agentic engineering, built to go fast. No Electron. No Tauri. 100% Rust. Manage all of your coding agents without friction. We're in Alpha: super.engineering

@shcallaway @brexHQ nice lmk if you meet any pirates

I got my payout from @brexHQ acquisition today! Now that I am filthy rich, I am shutting down Sazabi and retiring to the Caymans. Farewell everyone

@JohnPhamous beautiful, just this case feels off to me

we're improving the typography for the vercel docs to make it easier to read

@simoncorry absolute banger

If you're curious about what I've been up to lately... Just trying to solve a trillion dollar blindspot, no biggie. ramp.com/blog/trillion-…

@stevelauda_ agents often have to map between code and Figma's proprietary format in-context vs Paper is simply a DOM canvas which maps cleanly

Figma MCP is more token hungry than Paper MCP. Why?

@warpdotco what kind of background makes a good recruiter?

6/6 offers accepted in the last 2 weeks but, we are still hiring! We've got roles open for: / Growth Engineer / Product Engineer / Account Executive / Account Manager / Founding Recruiter (Generalist, GTM Focus) We're a small team taking on billion dollar incumbents, building the most powerful employee management platform for ambitious companies. If you want to move fast and ship constantly, you'll be in the right place.

@ayushsoni_io @Crestron ayo you guys hearing this?

imo interaction design + protyping is going to become the most valuable skill for designers. being able to come up with novel ways of interacting with AI + reducing friction is the layer that brings radical adoption to already existing tech.

@sonofalli patched maybe, but i don't think we can force genuine care where it doesn't exist

I know at least 4 people now who have been notified of a life altering diagnosis (cancer, etc.) via their physician’s office releasing their results on an online health platform with no follow up from a doctor/actual person via phone, appointment, etc. Very negative experience and seems easily fixable?

@enjojoyy bit of both depends who's asking

are you a generalist or do you just not have any skills you're actually proud of

@thdxr deprecadoed

@thdxr this opinion has been depreciated

so my gen-z coworkers i noticed they say words wrong all the time or they'll mix up 2 similar sounding words is this a thing