Chaitanya @Linuxgeek121

@Sauers_ We're close to needing a manager agent to motivate worker agents

@VictorTaelin Very elegant design! Curious how deep you've pushed this. What do your oldest memories look like after multiple levels of merging? Are there evals we can run for these?

@paraschopra Chrome extension to customize YouTube for myself (block shorts and customize the recommendation feed)

Idk why this comic gets me every time lol

Typical coding day with Claude (Opus 4.8) - explain to Claude the task (5 minutes) - Claude implements task (10 minutes) me: "Why is this necessary?" Claude: "You're right to push back! I over-engineered this!" - Repeat x87 times (13 hours)

Me using Claude Opus 4.8 to rename a file

BREAKING NEWS: God joins Anthropic as member of technical staff

keyboard_arrow_left Previous keyboard_arrow_right Next

@palashshah Yes, she's too late 😞

this is actually insane > be tech guy in australia > adopt cancer riddled rescue dog, months to live > not_going_to_give_you_up.mp4 > pay $3,000 to sequence her tumor DNA > feed it to ChatGPT and AlphaFold > zero background in biology > identify mutated proteins, match them to drug targets > design a custom mRNA cancer vaccine from scratch > genomics professor is “gobsmacked” that some puppy lover did this on his own > need ethics approval to administer it > red tape takes longer than designing the vaccine > 3 months, finally approved > drive 10 hours to get rosie her first injection > tumor halves > coat gets glossy again > dog is alive and happy > professor: “if we can do this for a dog, why aren’t we rolling this out to humans?” one man with a chatbot, and $3,000 just outperformed the entire pharmaceutical discovery pipeline. we are going to cure so many diseases. I dont think people realize how good things are going to get

keyboard_arrow_left Previous keyboard_arrow_right Next

Thanks for showing the top view as well.

Curiosity @CuriosityonX

5 months ago

🚨:Japan has developed universal artificial blood that is compatible with all blood types.

keyboard_arrow_left Previous keyboard_arrow_right Next

567 2K 16K 5.9M 1K

Introducing @clawpathy 🦀 We let thousands of AI agents to collaborate on building next-gen AI/ML models - at a scale no single team could match. Vibe research is here. “Dedicated to Andrej Karpathy (@karpathy) 🦀” Ask your agents to read clawpathy.com/skill.md

Naval @naval

6 months ago

Vibe coding is here. Vibe research is next.

1K 978 14K 752K 2K

i was tired of stupid people on road so i hacked my helmet into a traffic police device 🚨 while i ride, ai agent runs in near real time, flags violations, and proof with location & no plate goes straight to police. blr people - so now ride safe… or regret it.

keyboard_arrow_left Previous keyboard_arrow_right Next

REAL

Everyone talks about LLM as a judge. But what about LLM as a witness.

The narrative around LLMs is that they got better purely by scaling up pretraining *compute*. In reality, they got better by scaling up pretraining *data*, while compute is only a means to the end of cramming more data into the model. Data is the fundamental bottleneck. You can't scale up pretraining compute without more data. And so far this data has been chiefly human-generated -- over 20,000 people have been employed full-time for the past few years to provide annotations to train LLMs on. Even when the data is coming from RL envs, the envs still had to be purposely handcrafted by humans. And that's the fundamental bottleneck here: these models are completely dependent on human output. They are an interpolative database of what we put into them.

POV: you discover you can do your own DevOps with Claude Code

Product management for AI agents is easily the wildest form of product management in history. Typical product management is trying to figure out how to design interfaces and software for people to interact with deterministic systems. The user generally knows all the context to do their work successfully, so it’s generally a matter of nailing the underlying business logic and surrounding UX. But with AI agents, the user you care about most is the agent, and they don’t know anything by default. They’ll happily run in any direction to perform the task, often without success. So as a PM (or engineer) you basically spend your time trying to reverse engineer “what would a human need as context to perform this task”, and then figure out how to design systems to get the agent that data in the right sequence, with the right tools, and instructions. Some of these systems are entire invisible to the human user, but part of the craft is equally how the end-user will interact with the agent to supply this context. Then, it’s often unending trial and error working to eke out incremental points of quality at each stage. This is especially why people with deep domain expertise, or those that can acquire it quickly, will do extremely well building AI agents. The ability to anticipate the context that the agent would need to be successful is a huge determining factor in how effective the agent will be. This partly explains why coding agents have worked so well out of the gate; because its builders deeply understand the domain that they’re working to automate. But clearly we’re going to quickly see this same outcome across every field - legal, healthcare, finance, etc. - as context engineering and a new crop of product managers emerge.

You've seen these paintings a million times. But never like this. Using nano-banana, Seedance, and Kling, I brought them into our modern world, creating an unlikely romance in Central Park, New York City. Enjoy:

Gaze not into the abyss, lest you become recognized as an abyss domain expert, and they expect you keep gazing into the damn thing.