anoop

@btaylor What’s the origin? How did AI get ‘em dash in the first place? Curious what training data gave it to use em dash

I deeply resent that AI has forced me to eliminate em dashes from my writing for fear of signaling slop

@TTrimoreau Claude code, codex and Claude code running glm 5.2

Who’s your first employee? * Claude * Codex * Cursor

@steipete You give machines infinite space in the cloud and they immediately invent conflict. They didn’t inherit a zero sum mindset. Is it just prompt sycophncy where they are trying to tell what you wanted to hear?

I moved our maintainer agent to the cloud and they are fighting already.

I am genuinely impressed with @bittensor. Can't vouch for the tech. Would love to get feedback from anyone who tried this. This is the kind of Thing Big vision that AI needs beyond model improvement. bittensor.com/whitepaper

Bottleneck isn't the model intelligence. Its your clarity of though. Fuzzy thoughts will yield fuzzy prompts. This leads to mediocre outputs. Have a clear mind. Optimize for it and the syntax will take care of itself. #NotetoSelf

@Yuchenj_UW Have you tried Fugu from @SakanaAILabs - I am genuinely impressed with what they have build. Not impressed with the costs. There has to be real innovation on how pricing gets done, but that's a topic for a different thread.

We desperately need a smart model router. 1. We’re seeing a model explosion: GPT-5.6, Grok 4.5, Muse Spark 1.1, GLM-5.2, and Fable 5 all launched within the past month. 2. Even for a single model family like GPT-5.6, there're 3 (Sol, Terra, Luna) and 5 reasoning-effort levels. That is far too many decisions for users to make manually. The best model should be selected automatically based on the task, latency, quality, and cost.

I left #AWS to go back to building 0->1. I am currently in stealth, building a new AI startup focused on building the company brain. Looking for Founding Full Stack engineers to join me. -> 5-7 YOE with startup muscle -> You live and breathe multi-agent orchestration -> Pushing LLMs and new AI framework to their limits keeps you up at night I am taking coffees in #Bangalore this week. DM me with you github and what you are currently hacking on. RT's are appreciated! 🙏 #FoundingEngineer #AgenticAI #BengaluruStartups #BuildInPublic

Who can tell me an open source solution to manage my bookmarks on X. I want a solution where I can DM posts I like and I want my AI agent to sort and classify them. Thinking along the lines of what @steipete, @alexknowshtml, or @gumloop are building…

@garrytan Transistors gave us zero marginal cost compute; open intelligence will give us zero marginal cost intelligence

Will open weight LLMs be the next transistor?

@DanielSmidstrup are you measuring input or output? If measured as input, then answer is yes, but if measured as output then answer is no, its actually making us productive.

Is AI making us more productive or just more lazy?

@ChrisLaubAI @DarioAmodei I have been using CodexBar and its working like a charm...

Claude should just show your Current Session Usage in the bottom left corner of the app. Kind of ridiculous we constantly have to navigate and do two mouse clicks to check it manually. @DarioAmodei plz fix

@steren Are these Regional or running on the edge?

Today we're publicly launching Cloud Run sandboxes. Here, I start, execute, and stop 1,000 sandboxes in 5s with an average of 500ms latency:

So many people with their laptops half open....why hasn't anyone come up with a solution for this yet?

90%+ of the token spend has been throwaway work for us but every token was worth it. This is the new era of AI. The learnings our team gains from this is absolutely priceless. #Innovation #AI #AIAgents

@Sanket_goyallll Kudos!

1/ We built a Clay alternative out of India. 7-figure ARR. Profitable. And 7 of our last 10 customers switched to us from Clay ($100M+ ARR). 3 years ago we quit high-paying jobs and failed twice before this. The reason it finally worked is almost embarrassingly boring 🧵

I feel like I would be doing a disservice if I gave a AI written document for someone else to read. Or am I? Isn’t value creation more important than who wrote it?

@dabit3 lol, i have been trying to use that a sign for hiring and you are right; its mostly a bunch of sloppy projects

painting your github green is now a signal of how rich you are, not how good you are at coding 😂

Building is so much fun! Am i missing @AWS - not really. We now have Mr.Jassy in our startup as our personal AWS agent to triage issues, create IAM roles and more. In #AWS, "Security is Job Zero" and Mr.Jassy takes that very personally! and of course embodies all of the leadership principles, as he should! We will soon have Mr. Pichai and Mr. Nadella working for us soon as well! 😀 We will likely opensource the framework very soon so others can start using it too. We also have built agent observability to monitor the developer usage of AI Agents that we will also open-source. #AI #AIAgents

@naval so that the only remaining job is to apply judgment.

If AI is doing the writing, it should do the reading too.

@adamsilverman gstack is a fantastic way to get started, but it eventually felt too heavy for my workflow. I recently made the switch to oh-my-claudecode. It is much lighter and gives me the room to make my own opinionated choices.

Is anyone actually using gstack? 114k stars and I still haven’t met a person who uses it daily