•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 26–50 of 1145 mentions
-
• twitter 18h negativeAI HAS ADDED $33 TRILLION TO THE S&P 500 AND GROCERIES ARE STILL JUST AS EXPENSIVE @AonicClub, from @maxjmeier, sells everything at cost and uses AI agents for sourcing and operations, so there's no brand, distributor or retailer markup 24 cans of aonic fuel: $38.62 ($1.61 a can). nurri at costco: $55.18, 30% less wi… AI HAS ADDED $33 TRILLION TO THE S&P 500 AND GROCERIES ARE STILL JUST AS EXPENSIVE @AonicClub, from @maxjmeier, sells everything at cost and uses AI agents for sourcing and operations, so there's no brand, distributor or retailer markup 24 cans of aonic fuel: $38.62 ($1.61 a can). nurri at costco: $55.18, 30% less with aonic 24 servings of aonic fuel powder: $23.46. optimum nutrition: $57.63, per 24 servings of 32 g of protein, 59% lessview source ↗
-
• twitter 20h negativemost businesses overpay for AI by 10x. not because AI is expensive. because they use the most powerful model for everything. my rule for every build: 1. small, cheap model for sorting, extracting and formatting 2. big model only where real reasoning is needed 3. cache anything that repeats 4. check the bill every… most businesses overpay for AI by 10x. not because AI is expensive. because they use the most powerful model for everything. my rule for every build: 1. small, cheap model for sorting, extracting and formatting 2. big model only where real reasoning is needed 3. cache anything that repeats 4. check the bill every week, not every quarter the cheapest AI system isn't the one with the cheapest model. it's the one that uses each model for the right job.view source ↗
-
• twitter 20h negativeAlmost everyone starts with one AI agent and gives it everything. - A support ticket - A LinkedIn post - Something weird on the server - Then back to the support ticket Context windows are huge now, so why not have a single agent do it all? Well... more context doesn't mean better work. It's called context rot. The … Almost everyone starts with one AI agent and gives it everything. - A support ticket - A LinkedIn post - Something weird on the server - Then back to the support ticket Context windows are huge now, so why not have a single agent do it all? Well... more context doesn't mean better work. It's called context rot. The longer the input gets, the less reliable the model gets (even when the answer is right there). You'd never hire a single person to be your support team, your DevOps engineer AND your accountant. So don't do it with your agents either. Here's how we set ours up 👇view source ↗
-
• twitter 21h negative😭😭😭😭I made a major decision to scale down my ads from $4000/day adspend to just $2000/adspend (50% reduction) because 14 days of fucking October HELL I CHANGED IT AND I FINALLY PROFITED FUCKING BACK TO WHERE I NEED TO FUCKING ME 😭😭 FUCK SCALING I RATHER MAKE PROFIT THAN VANITY METRIC OF REVENUE CUNTS 😭😭😭😭I made a major decision to scale down my ads from $4000/day adspend to just $2000/adspend (50% reduction) because 14 days of fucking October HELL I CHANGED IT AND I FINALLY PROFITED FUCKING BACK TO WHERE I NEED TO FUCKING ME 😭😭 FUCK SCALING I RATHER MAKE PROFIT THAN VANITY METRIC OF REVENUE CUNTSview source ↗
-
• twitter 21h negativeWhat have they done to Owaisi ji? 😭🤣 Whoever did this! 😹 https://t.co/zc7uoTyEvY What have they done to Owaisi ji? 😭🤣 Whoever did this! 😹 https://t.co/zc7uoTyEvYview source ↗
-
• twitter 22h negativeI found a $15M app infra scrolling 😭 This app makes $400,000/mo and its so simple, all it does is record 40 second of UGC showing how the app works & run it at scale. i think even claude 1.0 could one shot this app.. a 1 feature app that plans meals for ordinary people is making $400,000/mo here's the or… I found a $15M app infra scrolling 😭 This app makes $400,000/mo and its so simple, all it does is record 40 second of UGC showing how the app works & run it at scale. i think even claude 1.0 could one shot this app.. a 1 feature app that plans meals for ordinary people is making $400,000/mo here's the organic UGC campaign they're running & can replicate: https://t.co/5rjJ34Bw06view source ↗
-
• twitter 22h negativeMuse Wonder & Wonder FULL by @AIatMeta on KingBench 4 Ranking: Opus 5.5 — 57.8% GPT-6 Astra — 56.5% GPT-6.1 Sol — 55.0% Haiku 5.5 — 54.2% Muse Wonder Full 1.0 — 53.0% Muse Wonder 1.0 and GLM-5.3 — 45.8% each, tied The models is not Opus 5.5 Level and is not token efficient at all. There are more downsides to the mode… Muse Wonder & Wonder FULL by @AIatMeta on KingBench 4 Ranking: Opus 5.5 — 57.8% GPT-6 Astra — 56.5% GPT-6.1 Sol — 55.0% Haiku 5.5 — 54.2% Muse Wonder Full 1.0 — 53.0% Muse Wonder 1.0 and GLM-5.3 — 45.8% each, tied The models is not Opus 5.5 Level and is not token efficient at all. There are more downsides to the model than there are upsides. GPT-6.1 Sol still remains the king at being token efficient. Sol costs about $1 per task, Haiku costs about $1.23 per task, Wonder costs about $1.75 per task. Wonder Full costs about $2.87 per task Astra costs $6.6 and Opus 5.5 - $7.5 This price could have been very justified if the performance was on level with Opus or Astra but it is below Sol and Haiku. Wonder is on GLM-5.3 level while Wonderfull is short of Sol & Haiku. Wonder is really bad at complex prompts, developing fully featured apps unattended. The bench for context tests for 6 prompts that involve building a docker clone (VM), GBA, DS Emulation, Room Desigining Software, Pokemon Red reconstruction, QMK/VIA Compatible Keyboard Configurator and so on. The model is not acing any of these tests. The full test results will be available on the YT Video: https://t.co/Z1XMPnAbILview source ↗
-
• twitter 22h negativeYo is there any plan to add a design mode in opencode any time soon? Like at least some type of annotations picker? Commandcode has one but I don't like their, let's say 'special' CEO at all and would rather use your harness. @thdxr @jayair @LukeParkerDev @opencode Yo is there any plan to add a design mode in opencode any time soon? Like at least some type of annotations picker? Commandcode has one but I don't like their, let's say 'special' CEO at all and would rather use your harness. @thdxr @jayair @LukeParkerDev @opencodeview source ↗
-
• twitter 23h negativeFrom my data, the worst hook in TikTok is the surprised face. 🙊 It looks scripted and got the lowest views for us Instead, when we tried darker-lighting, slanted camera angle and more subtle facial expressions (irritation, awkward laugh, sadness) it boosted views by 500x https://t.co/Zu0kmetsyk From my data, the worst hook in TikTok is the surprised face. 🙊 It looks scripted and got the lowest views for us Instead, when we tried darker-lighting, slanted camera angle and more subtle facial expressions (irritation, awkward laugh, sadness) it boosted views by 500x https://t.co/Zu0kmetsykview source ↗
-
• twitter 1d negativeHOLY SHIT FORBES REALLY DON’T MISS 😭 https://t.co/blBL0ys5x3 HOLY SHIT FORBES REALLY DON’T MISS 😭 https://t.co/blBL0ys5x3view source ↗
-
• twitter 1d negativeBreaking Bad star RJ Mitte shared tragic final post just hours before his death age 34 https://t.co/nXQgL8p7hH https://t.co/OVKrdlWYQZ Breaking Bad star RJ Mitte shared tragic final post just hours before his death age 34 https://t.co/nXQgL8p7hH https://t.co/OVKrdlWYQZview source ↗
-
• twitter 1d negative@shadcn Now that benchmarks mostly feel useless, I think this will be my focus going forward. Just finished filming the vid. Vast majority of it is talking about things like this :) @shadcn Now that benchmarks mostly feel useless, I think this will be my focus going forward. Just finished filming the vid. Vast majority of it is talking about things like this :)view source ↗
-
• twitter 1d negativesubmitted 𝗗𝗼𝗰𝗸𝗲𝘁 for the @monad Metropolis hackathon. agents are beginning to make payments, but giving them money can potentially lead to three issues: they can be prompt-injected, their keys can leak, and their judgment may not reflect yours. so i built Docket, an AI agent's reputation, computed from what a contrac… submitted 𝗗𝗼𝗰𝗸𝗲𝘁 for the @monad Metropolis hackathon. agents are beginning to make payments, but giving them money can potentially lead to three issues: they can be prompt-injected, their keys can leak, and their judgment may not reflect yours. so i built Docket, an AI agent's reputation, computed from what a contract actually let it do. Docket keeps the funds and policies in a contract, while agents receive limited permissions to act. even denied actions are recorded on-chain, creating a permanent trace of what an agent attempted, not just what it was allowed to do. the goal is to make agent activity more accountable. live URL: https://t.co/ha6ao9inBx source: https://t.co/EfcdqSBAvK @monad_devview source ↗
-
• twitter 1d negativejust hit $545k MRR... still the only full-time dev and the app is over 1m lines of code... in the last 7 days i merged 1,089 PRs here's the tech stack.. 1. over 50 oh-my-pi agents running in parallel on my home linux server 2. 8 Max Claude plans, 1 SuperGrok Heavy plan, and 5 max Codex plans 3. Slack & Telegram… just hit $545k MRR... still the only full-time dev and the app is over 1m lines of code... in the last 7 days i merged 1,089 PRs here's the tech stack.. 1. over 50 oh-my-pi agents running in parallel on my home linux server 2. 8 Max Claude plans, 1 SuperGrok Heavy plan, and 5 max Codex plans 3. Slack & Telegram to communicate with my bots. I no longer use a terminal or IDE 4. Linear, the central PM tool my bots use for issue tracking. Every engineering request, bug report, alert and error gets posted here 5. Supabase staging database + Railway preview environments for testing 6. Ubicloud runners for GitHub Actions 7. 4 Mac minis as self-hosted runners for iOS builds and 15 end-to-end iPhone journey tests 8. Sentry for error monitoring 9. pstack by @poteto, my main agent uses her orchestrate playbook to manage the other 50 sessions. it never touches code. 10. a custom 5-stage planning harness before any code gets written. 11. customized pr-af and codex for independent code reviews, so that the model that wrote the code never passes it 12. greptile for final code reviews 13. a watcher that tracks every web merge in production for 24 hours, and reverts it on its own if real users start failingview source ↗
-
• twitter 1d negativeBuild for one person first. The rest can follow. I built PoundKit to fix my own annoying problem. Turns out many others needed the same thing. Are you building something for your own pain points? https://t.co/fOzCir13ki Build for one person first. The rest can follow. I built PoundKit to fix my own annoying problem. Turns out many others needed the same thing. Are you building something for your own pain points? https://t.co/fOzCir13kiview source ↗
-
• twitter 1d negativeGrok Bot was given $48, told to stay out of the trash and turned it into $7,492 in 24 hours across 278 trades with 232 green, never needing a moonshot because the balance kept compounding under every big close. https://t.co/R33hp90uGA Grok Bot was given $48, told to stay out of the trash and turned it into $7,492 in 24 hours across 278 trades with 232 green, never needing a moonshot because the balance kept compounding under every big close. https://t.co/R33hp90uGAview source ↗
-
• twitter 1d negativeLaunching Nova on web today by @supercodeai an open-source alternative to @capydotai and @devindevelopers — an AI engineer for your development workflow. • agent threads to keep work organized • free pull request reviews • bug triage and proposed fixes • automations for recurring engineering work • gitHub, Linear, … Launching Nova on web today by @supercodeai an open-source alternative to @capydotai and @devindevelopers — an AI engineer for your development workflow. • agent threads to keep work organized • free pull request reviews • bug triage and proposed fixes • automations for recurring engineering work • gitHub, Linear, Slack & more, powered by @composio • cloud agents with configurable compute — 4 GB to 64 GB • deep search for AI researchers, powered by @getcontextdev and @firecrawl • BYOK: use your own models, with special benefits for @concentrateai users • connect your existing subscriptions — OpenCode, Claude, Codex & more — and use them inside Nova Now available on both web and desktop.view source ↗
-
• twitter 1d negativethis has to be a scam, right? https://t.co/bjM5N0Dhb7 this has to be a scam, right? https://t.co/bjM5N0Dhb7view source ↗
-
• twitter 1d negativeThink about the last time you put your glasses back in their case. Grip. Turn. Line it up. Let go. You didn't think about any of it. For a robot, that's one of the hardest problems there is. @naturewill Natura-1 learned it from nearly 1M hours of egocentric human data, and it handles tasks it was never trained on. … Think about the last time you put your glasses back in their case. Grip. Turn. Line it up. Let go. You didn't think about any of it. For a robot, that's one of the hardest problems there is. @naturewill Natura-1 learned it from nearly 1M hours of egocentric human data, and it handles tasks it was never trained on. Watch the hands 👇view source ↗
-
• twitter 1d negativeRobots can do backflips. Most still can't fold a towel. Nature Will @naturewill just launched NW-1, a foundation model built for dexterous hands. It goes straight at the problem every robotics lab keeps running into. It was trained on nearly 1 million hours of egocentric human data. First-person recordings of people'… Robots can do backflips. Most still can't fold a towel. Nature Will @naturewill just launched NW-1, a foundation model built for dexterous hands. It goes straight at the problem every robotics lab keeps running into. It was trained on nearly 1 million hours of egocentric human data. First-person recordings of people's hands doing everyday work, paired with the scene around them and what they were trying to do. So it doesn't memorize one task. It learns how humans actually handle objects. That shows up in 3 ways: 1. Zero-shot instructions. Give it a new destination, orientation, or constraint and it adapts the motion without fresh demos. 2. Dexterous control. Contact points shift, objects rotate, and two hands swap roles mid-task. It keeps control while the interaction unfolds. 3. Long-horizon execution. It tracks what's already been opened, moved, or attempted, and plans the next step from there. Watch the demos. Glasses go into a case. A towel gets unfolded. One hand holds an object while the other picks up a second. Boring tasks. That's the point. Your home is full of them, and none come with training data. The hard part of robotics was never strength or speed. It was always the hands. NW-1 is built around that problem. https://t.co/4zaRmnF4Ilview source ↗
-
• twitter 1d negativeBro cracked the code 😭😭 https://t.co/S95JiPSjhG Bro cracked the code 😭😭 https://t.co/S95JiPSjhGview source ↗
-
• twitter 1d negative🚨 with an X subscription, you get grok, cursor, and x premium. this is a much needed bundle for most of us. the only catch is that cursor does not let you use third party models. i use claude code for coding, so that is a dealbreaker for me. what do you guys think about this new subscription? 🚨 with an X subscription, you get grok, cursor, and x premium. this is a much needed bundle for most of us. the only catch is that cursor does not let you use third party models. i use claude code for coding, so that is a dealbreaker for me. what do you guys think about this new subscription?view source ↗
-
• twitter 1d negativeJust realized the "paid shill" accusation are kind of a self own. Like I'm sorry you're so broke that you're willing to lie for money. Couldn't be me :) Just realized the "paid shill" accusation are kind of a self own. Like I'm sorry you're so broke that you're willing to lie for money. Couldn't be me :)view source ↗
-
• twitter 1d negativeAsked Opus 5.5, GPT-6.1 Sol, and Muse Wonder Full to fix a bug Opus found the root cause and filed a pr GPT-6.1 Sol asked a question, crashed from the async question tool interrupt, found the bug, fixed it, wrote 300 lines of tests, and explained it in impossible to understand terms. Muse Wonder Full found a small P… Asked Opus 5.5, GPT-6.1 Sol, and Muse Wonder Full to fix a bug Opus found the root cause and filed a pr GPT-6.1 Sol asked a question, crashed from the async question tool interrupt, found the bug, fixed it, wrote 300 lines of tests, and explained it in impossible to understand terms. Muse Wonder Full found a small PR someone else had filed and said "you should probably merge this instead" I really like this model.view source ↗
-
• twitter 1d negative🚨MAS DUDAS🚨 RJ Mitte había revelado hace poco cómo fue explotado financieramente a los 17 años mientras intentaba proteger sus ganancias de Breaking Bad: “He pagado más en impuestos de lo que gané con Breaking Bad. Como un joven, tenía un gran equipo, viajé por el mundo y gasté dinero manteniendo mi carrera en marcha. … 🚨MAS DUDAS🚨 RJ Mitte había revelado hace poco cómo fue explotado financieramente a los 17 años mientras intentaba proteger sus ganancias de Breaking Bad: “He pagado más en impuestos de lo que gané con Breaking Bad. Como un joven, tenía un gran equipo, viajé por el mundo y gasté dinero manteniendo mi carrera en marcha. Todo volvía al negocio.” Un mal gerente de negocios me empeoró las cosas. “Durante los últimos cuatro años de mi serie de mayores ingresos, quería proteger mi dinero. Tenía 17 o 18 años, y todos me decían que consiguiera un contador adecuado.” Fui estafado. “Me devoraron vivo.” La contadora tenía otros 14 actores, incluyendo personas con discapacidades y actores mayores: ''Me escribía un cheque para pagar mis impuestos, y en lugar de pagarlos, ella me enviaba una declaración de impuestos falsa y se quedaba con el dinero.” Aún así, asume la responsabilidad de lo que pasó. “Al final del día, sin embargo, fue mi negligencia. Al IRS no le importas realmente.” QUED😔view source ↗