Anhad Jai Singh

Good list to see everything that shipped, but putting dispatch in D tier (while missing channels entirely?) and putting Computer use in C tier is criminal. They are both actual game changers. Auto mode should be nowhere close to S tier.

@cullenroche So, uh, can I ask for your take instead then?

Live shot of me avoiding every Iran take on this website.

how do u fix openclaw internal reasoning leaking

@TheStalwart GPT 4o still out here being used in the wild

I don’t mind the AI reply bots. Some of the last sane and sober voices on here

@TheStalwart @tracyalloway @heyitsnoah Was listening to the pod and one thing that stood out to me was the confusion between Tracy's question about memory notes (something chatGPT web does) and file writing for Claude code and compaction. I feel like you guys should actually talk to @bcherny for way more deets!

NEW ODD LOTS: Claude Code episode @tracyalloway and I talk to @heyitsnoah, who’s been using LLMs since before ChatGPT existed, about what’s different about Claude Code, why everyone’s going crazy for it, and the big threat to legacy software companies podcasts.apple.com/us/podcast/odd…

Claude Code with Opus 4.5 is a watershed moment, moving software creation from an artisanal, craftsman activity to a true industrial process. It’s the Gutenberg press. The sewing machine. The photo camera.

@geoffreylitt "I gen-ai'd it" is what I usually say, sometimes I shorten it to "I gen'd it"

We need a shorthand way of saying: "An AI did the work, but I vouch for the result" Saying "I did it" feels slightly sketchy, but saying "Claude did it" feels like avoiding responsibility

@NateSilver538 @TheStalwart FWIW - I think these are active decisions made inside the big frontier model labs all the time - datasets for things like Coding, Math, Music, Writing, Quant skills are high ROI to add. biz/co will pay for them. Poker, Chess, Pokemon etc aren't, but you could be pretty easily.

@NateSilver538 @TheStalwart I have a different take on this - LLMs are smart over what they've seen before. Poker/chess just aren't dedicated datasets for training frontier models yet. If someone took the corpus of all hist games + annot. for a good/bad plays, fed it into pre-training, perf would 🚀

In what year will an LLM beat a GM at chess?

Can one of the VCs in SF pls fund an internet company that provides good Internet and half decent service? AT&T and Xfinity cannot be the best of what Silicon Valley has to offer. It's not venture scale; it won't make $$$, but pls do it anyway, just for for the love of the game.

What do you folks use to organize knowledge in 2025? I consume a bunch of media - blogs, podcasts, news articles, tweets/threads etc etc. Tell me what systems/tools work for you - anything from handwritten notes, to a personal obsidian, NotebookLLM - anything you actually use

Terminal rendering magic is very archaic and probably one of the less understood/used/loved for UI systems today - I find it a bit funny, but if not for claude code it would likely have stayed that way.

@maxrumpf @Zoom > Maybe someone on Wall Street will believe it. Not after this tweet

This is such a sad showing from Zoom. A note from someone who has *trained* a SOTA LLM. Let me explain: @Zoom strung together API calls to Gemini, GPT, Claude et al. and slightly improved on a benchmark that delivers no value for their customers. They then claim SOTA. The crime here is not using the models that are best at their tasks. This is actually quite smart and most applications should do this. @SierraPlatform uses multiple models although their CEO sits on OpenAI's board. Quite a strong endorsement for the technique. The crime is that their claim is hollow. They did none of the work. They did not train the model, but obfuscate this fact in the tweet. The injustice of taking credit for the work of others sits deeply with people. The "sad" part is that Zoom could train a SOTA LLM for something their users care about. Retrieval over call transcripts is not "solved" by SOTA LLMs (I know this for a fact, because we have an RL env for this @SID_AI). I figure Zoom's users would care about this much more than HLE. But this is not an attempt to help their users. They want the status of being a lab (and maybe the valuation that goes along with this). Maybe someone on Wall Street will believe it.

genuinely innovative pivot in contrast with the we build X, but ".ai" startup climate

Lando Norris and Oscar Piastri have been disqualified from the #LasVegasGP 🇺🇸

#FIA #F1

🔥 Today we’re excited to announce new funding for `grep` (at a $1.3B valuation) to continue building the foundation of agent observability and text search infrastructure. grep began as a humble UNIX utility in 1973. Since then, it’s evolved—through recursive innovation and the rise of ripgrep—into a core platform for developers, sysadmins, and agents. Our tools now power engineering and AI teams across @OpenAI, @Anthropic, @Meta, @Cloudflare, @Replit, @NASA, and thousands more. Over the decades we’ve iterated from grep to `egrep` to `ripgrep`. Our goal has always been to figure out what intelligent agents of the future need to see, filter, and extract—and then build the tools that make that possible. While our journey is still just beginning, we also want to take a moment to reflect on how the space (and our role in it) has evolved. You can read our reflections and details on this funding milestone here: gnu.org/software/grep/… We also share more about the funding that will power our future there. Thank you to @IVP, @Benchmark, @Sequoia, @CapitalG, and the open-source community for their belief in the enduring power of regex. What excites us most today is what’s next: grep 5.0 with AI-assisted pattern synthesis ripgrep Cloud, bringing distributed search to agent clusters pgrepGPT, an agent-native process discovery layer And new no-code integrations for autonomous observability pipelines We’re in the midst of a transformation in computation itself. grep and ripgrep will remain at the core—helping humans and agents alike find what matters, faster.

@signulll No, if you do that - you announce to every enterprise customer that they will lose their business suite if they compete with Google in any market it cares about, there's no faster way to send customers running. The GSuite doesn't have the leverage, there's Oficce356 etc.

if you were google, would you cut off openai's access to gmail, calendar, docs, & slides api's? a more aggressive ceo would've done it already.

@GergelyOrosz Yes, but Grafana also sucks - it's what everyone is forced to use when they can't get something better. You need your observability tools to be mostly turn key and grafana just isn't.

A product that (almost) everyone uses from mid-sized tech companies and up but I rarely hear talked about: Grafana In The Pragmatic Engineer 2025 survey, it had more mentions than Cursor, and dominates as the answer to "how do you turn information into graphs" This is Grafana:

ChatGPT (web) rendering for math notation borked? cc: @OpenAI / @ChatGPTapp

Despite the Gemini team training an excellent frontier model, I can't help but feel that Google is lacking direction on the AI product side. Why can't I give Gemini agentic access to my Gmail? Why can't it read my Google Docs to get the context it needs? Why can't I ask it to do data analysis in a Google Sheet? I should be able to ask Gemini to plan my vacation for me and it should be able to buy a flight on Google Flights (credit card info saved in Google Wallet), look up directions on Google Maps, cross-reference with my travel history via Gmail, and populate a Google Sheet with an itinerary. Unfortunately, Google has completely dropped the ball on giving Gemini the tools and context it needs to actually be useful. Instead, we get lazy, disconnected AI features like "help me write" or "chat with AI" in every single product. I truly feel that Google could do something special as the only AI company that natively has full context on the user's life. But for now it seems they lack the top-down product vision to make something great here.