•twitter15877
listening to
timeline
15877
mentions
tracked
tracked
net sentiment
+24%
4998 positive
9734 neutral
1145 negative
volume & sentiment over time
Sep 30 – Oct 14, 2026 · Daily
sources
most discussed
—
active days
15
mentions
Showing 2851–2875 of 4998 mentions
-
• twitter 1w positiveAre people actually being serious right now? People are losing their minds over a basic PC build just because it has an older RTX 4000 with 24GB VRAM inside. What makes this better than a 128GB DGX Spark? Or even an AI Max, which costs less than that whole setup? The replies are full of pure overhype calling this "t… Are people actually being serious right now? People are losing their minds over a basic PC build just because it has an older RTX 4000 with 24GB VRAM inside. What makes this better than a 128GB DGX Spark? Or even an AI Max, which costs less than that whole setup? The replies are full of pure overhype calling this "the future of local AI." Be very careful with these people doing paid hype posts just to make a quick buck and disappear.view source ↗
-
• twitter 1w positiveDoing some research and could really use your input: 1- If you could only keep one model and ditch the rest, what are you picking? 2- If it has to be open source, what's your pick? 3- For whatever you chose, what's the biggest thing it's missing? Appreciate it. Doing some research and could really use your input: 1- If you could only keep one model and ditch the rest, what are you picking? 2- If it has to be open source, what's your pick? 3- For whatever you chose, what's the biggest thing it's missing? Appreciate it.view source ↗
-
• twitter 1w positiveI MADE A MINECRAFT MOD WITH CLAUDE you shoot a hook into a tree and it pulls you up then you just keep swinging from one tree to the next the first version kept throwing me into the ground lol took me a few evenings to fix that flying over the forest at sunset was worth it https://t.co/VaHb9Mp0oE I MADE A MINECRAFT MOD WITH CLAUDE you shoot a hook into a tree and it pulls you up then you just keep swinging from one tree to the next the first version kept throwing me into the ground lol took me a few evenings to fix that flying over the forest at sunset was worth it https://t.co/VaHb9Mp0oEview source ↗
-
• twitter 1w positiveA new open-weights model is always good news. Even better when it is backed by Nvidia. But why do new AI labs always compare their models to the previous generation instead of the latest one? They compared against GLM 5.2 instead of GLM 5.3. A new open-weights model is always good news. Even better when it is backed by Nvidia. But why do new AI labs always compare their models to the previous generation instead of the latest one? They compared against GLM 5.2 instead of GLM 5.3.view source ↗
-
• twitter 1w positiveThis Pi release only took 4 hours. GitHub actions is amazing. This Pi release only took 4 hours. GitHub actions is amazing.view source ↗
-
• twitter 1w positiveLife is not about being rich, being popular, being highly educated or being perfect. It's about being real, humble and kind. Life is not about being rich, being popular, being highly educated or being perfect. It's about being real, humble and kind.view source ↗
-
• twitter 1w positivei knew that humongous android tablet of mine will eventually pay off after ... 4 years. this is really nice to work with. agents run on my phone, tablet connects to it. https://t.co/vtvQaLxCMJ i knew that humongous android tablet of mine will eventually pay off after ... 4 years. this is really nice to work with. agents run on my phone, tablet connects to it. https://t.co/vtvQaLxCMJview source ↗
-
• twitter 1w positive❤️❤️🙌❤️❤️ https://t.co/rdGl1KmNht ❤️❤️🙌❤️❤️ https://t.co/rdGl1KmNhtview source ↗
-
• twitter 1w positiveFor me it ends up being more like: *Tell AI: "I really want to make sure I get this all done, I'm definitely going to make sure it gets done" AI: "Here's a 5 page strategy for making sure you stick to it" *doesn't read it, just saves a copy somewhere on my computer* *50 new chats opened with to-do list items and id… For me it ends up being more like: *Tell AI: "I really want to make sure I get this all done, I'm definitely going to make sure it gets done" AI: "Here's a 5 page strategy for making sure you stick to it" *doesn't read it, just saves a copy somewhere on my computer* *50 new chats opened with to-do list items and ideas later, I'm just waving goodbye to whatever the original list was as it recedes into the horizonview source ↗
-
• twitter 1w positiveSend this to the teammate who keeps asking "how does Pi actually work?" 100 pages 26 diagrams Written from the source, not the README Agent loop, sessions, compaction, extensions, MCP, codemode, and the experimental Durable stack Free PDF: https://t.co/bodbMN8aWG Which repo should get one next? Send this to the teammate who keeps asking "how does Pi actually work?" 100 pages 26 diagrams Written from the source, not the README Agent loop, sessions, compaction, extensions, MCP, codemode, and the experimental Durable stack Free PDF: https://t.co/bodbMN8aWG Which repo should get one next?view source ↗
-
• twitter 1w positiveOpenAI might have a serious retention problem. Opus 5.5 is ridiculously good and costs far less. Fable 5.5 is about to make things even harder. Even a stronger Astra tomorrow won’t matter much if users run out of usage in a day. OpenAI might have a serious retention problem. Opus 5.5 is ridiculously good and costs far less. Fable 5.5 is about to make things even harder. Even a stronger Astra tomorrow won’t matter much if users run out of usage in a day.view source ↗
-
• twitter 1w positive＼グーコロ男爵発売記念！／ キャンペーンに参加すると抽選で景品がもらえます♪ ① @akiko_lawson をフォロー✨ ②「ローソンのグーコロ男爵」＆コメントを入れて引用ポスト🔁 ③ 抽選で1名様にQUOカードPay10,000円分プレゼント🎁 ＼グーコロ男爵発売記念！／ キャンペーンに参加すると抽選で景品がもらえます♪ ① @akiko_lawson をフォロー✨ ②「ローソンのグーコロ男爵」＆コメントを入れて引用ポスト🔁 ③ 抽選で1名様にQUOカードPay10,000円分プレゼント🎁view source ↗
-
• twitter 1w positivebuilt my own push-to-talk voice assistant over the weekend to try elevenlabs v4! Scribe v2 → Sonnet 5.5 → ElevenLabs v4, on Cloudflare Workers, with a little harness for tools + memory. sure grok bot and dots are cool, but building your own hits different. video made by Opus 5.5 (also insane!) sound on built my own push-to-talk voice assistant over the weekend to try elevenlabs v4! Scribe v2 → Sonnet 5.5 → ElevenLabs v4, on Cloudflare Workers, with a little harness for tools + memory. sure grok bot and dots are cool, but building your own hits different. video made by Opus 5.5 (also insane!) sound onview source ↗
-
• twitter 1w positiveIntroducing omni-macos: a semantic Finder for Apple Silicon. It was first released months ago, but now getting really good and becoming my daily productivity. Over the last few months I've kept improving fully local, on-device semantic search for my everyday agentic use. The goal: a local multimodal search engine that… Introducing omni-macos: a semantic Finder for Apple Silicon. It was first released months ago, but now getting really good and becoming my daily productivity. Over the last few months I've kept improving fully local, on-device semantic search for my everyday agentic use. The goal: a local multimodal search engine that's easy for both humans and agents, fast, and handles CRUD in real time. After years of working on embedding and reranker models, I figured this would be easy, especially with coding agents hill-climbing tps for you. Turns out making it actually good is pretty damn hard. I've also seen other projects like zvec-grep going after the same thing, but I think most of them miss some key points about local search, especially on personal devices. They optimize for efficiency and relevance, which is fair with a SaaS mindset, but not enough for an air-gapped personal device: - Efficiency matters, but on a laptop, an indexer that hogs the machine to finish sooner is worse than one that takes a bit longer and you never notice. - Personal files are massively duplicated: v1, v2, v-final, v-final-final, plus backups, often with only tiny diffs between them. So using the file as the indexing unit is inefficient. - Some files get edited all the time, but only a small part changes each time. Reindexing the whole file is, again, a waste. - CRUD needs to be handled properly at the folder level. - You only have one GPU interface and it's non-preemptive. Embedding needs it, vector search needs it, OCR needs it. You can't lower the priority of work that's already queued, so priority can only be set at admission.view source ↗
-
• twitter 1w positiveIn 2007 buffett's partner told a room of law grads the habit that beats IQ, and almost nobody watches past the jokes charlie munger's USC law school speech, free on youtube those graduates paid well over $100,000 in tuition, and the most useful part came free on the last day munger was warren buffett's partner at be… In 2007 buffett's partner told a room of law grads the habit that beats IQ, and almost nobody watches past the jokes charlie munger's USC law school speech, free on youtube those graduates paid well over $100,000 in tuition, and the most useful part came free on the last day munger was warren buffett's partner at berkshire hathaway for decades, the man buffett credited with pushing him from cheap stocks toward great businesses most people know munger from one-liners in short clips, never from the full speech they came from and none of it costs money the whole thing, for free: 1. be a learning machine, people rise who are "not the smartest" but go to bed "a little wiser" 2. read like buffett, who by munger's account spends half his time "sitting on his ass and reading" 3. spend the other half talking one on one with highly gifted people 4. cut envy, resentment and self-pity none of these needs talent or money, which is exactly why most people skip them number 1 is the one that compounds, a little wiser every night adds up the same way money does, and munger got almost a century of it at 20, with decades ahead of you, that habit is worth more than any stock tip you'll ever get his words, "the safest way to try to get what you want is to try to deserve what you want" people pay for courses on how to get ahead, and berkshire's other half said the safest shortcut is to deserve it Nobody picks their IQ, but everybody picks how they spend their evenings.view source ↗
-
• twitter 1w positive400g skyr + 24g Vivani 92% chocolate + lemon juice 44g + 2g protein = 46g protein Dark chocolate has healthy flavanoids and barely any sugar Great for your heart and brain! https://t.co/xtbZzP7ViA 400g skyr + 24g Vivani 92% chocolate + lemon juice 44g + 2g protein = 46g protein Dark chocolate has healthy flavanoids and barely any sugar Great for your heart and brain! https://t.co/xtbZzP7ViAview source ↗
-
• twitter 1w positiveMet the absolute legend @stevewoz 🍎 Thanks for making computing personal, Woz :) Now onward to personal intelligence. The story continues. https://t.co/CkN3bKkPia Met the absolute legend @stevewoz 🍎 Thanks for making computing personal, Woz :) Now onward to personal intelligence. The story continues. https://t.co/CkN3bKkPiaview source ↗
-
• twitter 1w positivethere were quite a few things that annoyed me about my initial version; some i fixed for the youtube version but *this* missed occasion for a nice transition drove me crazy. i had to fix it (before/after in the video) https://t.co/u9vJI6F8Rs there were quite a few things that annoyed me about my initial version; some i fixed for the youtube version but *this* missed occasion for a nice transition drove me crazy. i had to fix it (before/after in the video) https://t.co/u9vJI6F8Rsview source ↗
-
• twitter 1w positivePeople keep asking me if learning to code is still worth it A few years ago, my answer was an easy yes... today, I’m not so sure I’ve been building software for 15 years across startups, banks, retail, and large companies. I was CTO of two startups and had to learn infrastructure, databases, ETLs, scaling, observabil… People keep asking me if learning to code is still worth it A few years ago, my answer was an easy yes... today, I’m not so sure I’ve been building software for 15 years across startups, banks, retail, and large companies. I was CTO of two startups and had to learn infrastructure, databases, ETLs, scaling, observability, cloud, and everything that breaks when real users arrive. That experience makes building with AI almost unfair I usually know what needs to be built, what will probably break, and where to look when it does But here’s the uncomfortable part: Someone starting today can prompt an AI, spin up services, and deploy a real app in minutes Things that took us weeks or months Technology is moving too fast to tell people they need 10–15 years of experience before they can build serious software The biggest advantage might actually belong to those of us who built startups before LLMs We already had to ship fast with limited resources, we just had to do it without AI We know where systems break We know what matters And now AI lets us compress years of execution into hours I used to tell everyone, learn programming fundamentals, deploy servers, read logs, understand databases Today? I’m genuinely not sure that’s still the best adviceview source ↗
-
• twitter 1w positiveOne day, someone will love you in a way that makes you feel wanted and valued every single day. One day, someone will love you in a way that makes you feel wanted and valued every single day.view source ↗
-
• twitter 1w positiveQuick tl;dr on the Cowork change, because a lot of people were caught off guard by it. What changed: - old Cowork ran Claude's work inside a virtual computer on your own laptop - new Cowork runs that virtual computer in the cloud instead - each session gets its own sandbox (a private, walled-off space), and it gets de… Quick tl;dr on the Cowork change, because a lot of people were caught off guard by it. What changed: - old Cowork ran Claude's work inside a virtual computer on your own laptop - new Cowork runs that virtual computer in the cloud instead - each session gets its own sandbox (a private, walled-off space), and it gets deleted when the session ends What it means for you: - your laptop stops doing the heavy lifting, so you should see less battery drain, less disk space used, and better speed - closing your laptop doesn't stop the work anymore, so long tasks can keep running - you can use Cowork from your phone What stayed the same: - Claude can still only see files in the folders you added to the session. When it needs one of those files, the desktop app on your computer hands it over. Thanks to the Cowork team for explaining all of this in public. They said they'll keep making changes based on feedback, so if something feels off, tell them.view source ↗
-
• twitter 1w positiveI just found herdr-gpui, and wow, it's good. I highly recommend trying it out! https://t.co/Ke7U4YLXGh https://t.co/1FQ8NpQU8O I just found herdr-gpui, and wow, it's good. I highly recommend trying it out! https://t.co/Ke7U4YLXGh https://t.co/1FQ8NpQU8Oview source ↗
-
• twitter 1w positiveThese orb animations are everywhere right now and somebody just open-sourced them It’s called Thinking Orbs: nine dotted orbs, one for each state your product goes through working, searching, solving, listening, connecting, weaving, composing, breathing, shaping. Why I keep reaching for them: - they replace the gene… These orb animations are everywhere right now and somebody just open-sourced them It’s called Thinking Orbs: nine dotted orbs, one for each state your product goes through working, searching, solving, listening, connecting, weaving, composing, breathing, shaping. Why I keep reaching for them: - they replace the generic spinner in any “waiting” state and actually say what’s happening - two tuned sizes (64 for chat, 20 for inline text), auto light/dark from your page - plain 2D canvas no WebGL, no filters so nothing slows a page down - respect prefers-reduced-motion out of the box (static frame, still themed) - MIT ship them in a paid product without asking anyone One npm install, one component, one prop.view source ↗
-
• twitter 1w positiveHottest AI startups in San Francisco right now 🔥 1/ OpenAI - ChatGPT, GPT models, agents · @OpenAI 2/ Anthropic - Claude + Claude Code · @claudeai 3/ Instinct - personal AI agent that acts for you 4/ Cognition - Devin, the autonomous software engineer 5/ Perplexity - AI answer engine + agents 6/ Sierra - enterprise CX… Hottest AI startups in San Francisco right now 🔥 1/ OpenAI - ChatGPT, GPT models, agents · @OpenAI 2/ Anthropic - Claude + Claude Code · @claudeai 3/ Instinct - personal AI agent that acts for you 4/ Cognition - Devin, the autonomous software engineer 5/ Perplexity - AI answer engine + agents 6/ Sierra - enterprise CX agents 7/ Harvey - AI for legal · @harvey 8/ Thinking Machines Lab - Mira Murati's lab 9/ Mercor - expert data + evals for frontier labs 10/ Figure AI - humanoid robots 11/ SSI - Ilya Sutskever's safe superintelligence lab 12/ Etched - custom AI inference hardware 13/ Physical Intelligence - robot foundation models 14/ World Labs - Fei-Fei Li's world models 15/ Decagon - enterprise customer agents 16/ Glean - enterprise AI search · @glean 17/ Factory - agent-native software delivery 18/ Scale AI - data, evals, frontier systems 19/ HappyRobot - AI agents for enterprise ops 20/ Halluminate - AI training environments (just raised) SF is eating the world again. Bookmark this + follow the ones building in public.view source ↗
-
• twitter 1w positivei think the most exciting model release by a new western lab in recent months (behind jev, of course) is persimmon. people are massively underrating the utility of having great realistic user simulators. i would imagine they are quietly printing as a private API used in RL envs i think the most exciting model release by a new western lab in recent months (behind jev, of course) is persimmon. people are massively underrating the utility of having great realistic user simulators. i would imagine they are quietly printing as a private API used in RL envsview source ↗