•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 2826–2850 of 4998 mentions
-
• twitter 1w positiveThank you @AlexAndBooks_ and for all you do to champion reading and books, learning and life enrichment! Thank you @AlexAndBooks_ and for all you do to champion reading and books, learning and life enrichment!view source ↗
-
• twitter 1w positive.@0xSigil reveals testing frontier AI's cyber capabilities scared him enough to start moving his personal data off data centers and onto his own devices: "There is going to be a radical reformation of the economy as businesses and the economy shift to serving agents as one of its newest users on the internet, especial… .@0xSigil reveals testing frontier AI's cyber capabilities scared him enough to start moving his personal data off data centers and onto his own devices: "There is going to be a radical reformation of the economy as businesses and the economy shift to serving agents as one of its newest users on the internet, especially as humans trust AIs more to act on their behalf." "We were working on agents that could run on long horizon harnesses, and we were also testing different frontier models, and it was just really scarily apparent just how good at cyber they were." "It was really clear to me that, oh my God, Mythos class models are open source now. I think that really led to me wanting to build Underdog for myself." "I wanna bring my data onto my devices, off of data centers, which are some of the biggest targets." @underdogdotai @ConwayResearchview source ↗
-
• twitter 1w positiveNano Banana 2 is so much better than GPT Image 2.5 it's kind of insane. Same exact prompt. Guess which is Gpt vs Nano Banana below 👇 https://t.co/64Uux6vg3Y Nano Banana 2 is so much better than GPT Image 2.5 it's kind of insane. Same exact prompt. Guess which is Gpt vs Nano Banana below 👇 https://t.co/64Uux6vg3Yview source ↗
-
• twitter 1w positiveI promise …. once you realize you’re the one who can help you and it starts with you … life gets good 👍🏼 … pass this on to 3 people who need it or tag @ them in the comments https://t.co/s2qKmKLHyc I promise …. once you realize you’re the one who can help you and it starts with you … life gets good 👍🏼 … pass this on to 3 people who need it or tag @ them in the comments https://t.co/s2qKmKLHycview source ↗
-
• twitter 1w positivefree and open source reverse engineered adobe products https://t.co/hKAGzTdea0 https://t.co/0zbNfiBi4W free and open source reverse engineered adobe products https://t.co/hKAGzTdea0 https://t.co/0zbNfiBi4Wview source ↗
-
• twitter 1w positiveI’m 35. • A husband. • A dad. • A 24/7 tech lead while touching grass. • Running my autonomous AI meta harness wrapper on the night shift - catching bugs and shipping SaaS/PaaS while I sleep. I’m definitely not sharing the full journey while chasing $75K MRR, but as much as you need to get on that path yourself with … I’m 35. • A husband. • A dad. • A 24/7 tech lead while touching grass. • Running my autonomous AI meta harness wrapper on the night shift - catching bugs and shipping SaaS/PaaS while I sleep. I’m definitely not sharing the full journey while chasing $75K MRR, but as much as you need to get on that path yourself with a higher chance of success. In this AI/SI era, I want to #connect with people who are actually trying to ship: • 🛠️ Founders & indie hackers putting things out there • 🤖 AI engineers building working technical solutions • 🛡️ AI security folks keeping systems safe • 💡 Visionaries & game devs building cool stuff • 📈 Growth folks figuring out distribution • 🚀 VCs & investors backing builders Most things fail before one finally clicks. That’s just how this game works. If you’re on any of these paths, say hi 👋 and let’s help each other grasp possibilites of cutting edge tech. Drop below your current primary AI tools and how they help you on a high level.view source ↗
-
• twitter 1w positivepeople often ask how we're better than claude well: "It basically got to the same recommendations that I had gotten to with Claude in like a month in almost 30 minutes. Like, it was incredible. My mind is blown for sure." - Nirav Suraiya, founder of Teal & Tangent Ventures people often ask how we're better than claude well: "It basically got to the same recommendations that I had gotten to with Claude in like a month in almost 30 minutes. Like, it was incredible. My mind is blown for sure." - Nirav Suraiya, founder of Teal & Tangent Venturesview source ↗
-
• twitter 1w positiveRT @poteto: @kr0der @t3dotcodes love t3code! Projects in cursor are just too good though. you don't need to see threads cluttering up your… RT @poteto: @kr0der @t3dotcodes love t3code! Projects in cursor are just too good though. you don't need to see threads cluttering up your…view source ↗
-
• twitter 1w positiveWe are finally seeing what @reflection_ai has been cooking. Their latest model Beam looks to be competitive with other open models (though Qwen 3.8 Max still takes the lead). The lack of a competitive American open-source AI model is one of the biggest risks to U.S. national security, and the release of Beam is a cruc… We are finally seeing what @reflection_ai has been cooking. Their latest model Beam looks to be competitive with other open models (though Qwen 3.8 Max still takes the lead). The lack of a competitive American open-source AI model is one of the biggest risks to U.S. national security, and the release of Beam is a crucial step for the American AI ecosystem. As we laid out in our recently published "cyberdefense strategy for superintelligence", building a competitive American open model is critical to combat the possibility of AI sabotage (call to actions below). Why? An open-source model developer already controls every step of their training pipeline. They can potentially build a sleeper agent into their model and publish it online for anyone to download. This is not theoretical. >Sleeper agents have been demonstrated through weight tampering, data poisoning, emergent misalignment, and direct training. >They are incredibly hard to remove (Anthropic tried using reinforcement learning, adversarial training, and safety tuning -- but it only made the backdoor harder to detect). >There is currently no reliable way to examine the weights of an open model and confidently rule out the presence of a backdoor or sleeper agent. The fact that a model is open source does not make it safe and, until we have reliable ways of detecting sabotaged or backdoored models, we cannot truly trust a model. The only solution right now is to build our own open models whose training pipelines we trust. We are embarking on a mission to secure the world ahead of the oncoming wave of autonomous cyberattacks. Most people don't know what's coming, it's quite sobering when you realize that most orgs just aren't ready. All bugs must be found, systems need to be patched, and this work has to happen fast - with AI. But the AI patching and red-teaming all of our systems can be sabotaged. We have to be able to trust the systems building world security. The release of Beam is an important stepview source ↗
-
• twitter 1w positive20k more users is wild growth 🚀 20k more users is wild growth 🚀view source ↗
-
• twitter 1w positiveUgh, Google - why can't I connect custom MCP servers on an individual user account? Also it looks like "the web" didn't yet pick up that Gemini CLI turned into Antigravity CLI, which says a lot. https://t.co/Ib9yfg75hY Ugh, Google - why can't I connect custom MCP servers on an individual user account? Also it looks like "the web" didn't yet pick up that Gemini CLI turned into Antigravity CLI, which says a lot. https://t.co/Ib9yfg75hYview source ↗
-
• twitter 1w positiveWhat is the ‘King Midas’ problem for AI alignment? “King Midas specified his objective very clearly. Everything I touch should turn to gold. What could go wrong?” — @UCBerkeley Computer Science Professor Stuart Russell 🚀 Watch the full episode of AI Deep Dive: https://t.co/Vt8X9fSwKD What is the ‘King Midas’ problem for AI alignment? “King Midas specified his objective very clearly. Everything I touch should turn to gold. What could go wrong?” — @UCBerkeley Computer Science Professor Stuart Russell 🚀 Watch the full episode of AI Deep Dive: https://t.co/Vt8X9fSwKDview source ↗
-
• twitter 1w positiveGrokBot recently dropped 4 new features that I found incredibly useful in my business... and not enough people are talking about them. 00:00 Intro 00:20 Primary / Main Bot 07:11 Team Bot 14:30 Inline Mini Apps for Slack and Gmail 17:40 Bypass Bot Blockers 19:37 Bonus Tip https://t.co/nudjttSgSf GrokBot recently dropped 4 new features that I found incredibly useful in my business... and not enough people are talking about them. 00:00 Intro 00:20 Primary / Main Bot 07:11 Team Bot 14:30 Inline Mini Apps for Slack and Gmail 17:40 Bypass Bot Blockers 19:37 Bonus Tip https://t.co/nudjttSgSfview source ↗
-
• twitter 1w positiveOvertown, thank you for welcoming me on Saturday. These conversations matter to me. I care deeply about the future of Florida’s Black community; as a father, a neighbor, and someone who shares in that future. I hear your concerns, and I’m committed to addressing them by lowering costs, expanding opportunity, and keepi… Overtown, thank you for welcoming me on Saturday. These conversations matter to me. I care deeply about the future of Florida’s Black community; as a father, a neighbor, and someone who shares in that future. I hear your concerns, and I’m committed to addressing them by lowering costs, expanding opportunity, and keeping my promises. I’m grateful for the chance to earn your vote.view source ↗
-
• twitter 1w positiveTech take: the most underrated junior skill is reading code, not writing it. After every Codewars kata I study other people's solutions for 20 minutes. You learn more from one great one-liner than from ten solved-by-accident attempts Tech take: the most underrated junior skill is reading code, not writing it. After every Codewars kata I study other people's solutions for 20 minutes. You learn more from one great one-liner than from ten solved-by-accident attemptsview source ↗
-
• twitter 1w positiveReflection just dropped Beam, a 501B open-weight MoE with only 23B active parameters. Trained on nearly 24T tokens and 10,500 GB300s, it scores 80.9 on SWE-bench Verified and 80.1 on Terminal Bench 2.1. And they’re claiming 3–4x better inference efficiency than GLM 5.2. https://t.co/5BRlCurhAf Reflection just dropped Beam, a 501B open-weight MoE with only 23B active parameters. Trained on nearly 24T tokens and 10,500 GB300s, it scores 80.9 on SWE-bench Verified and 80.1 on Terminal Bench 2.1. And they’re claiming 3–4x better inference efficiency than GLM 5.2. https://t.co/5BRlCurhAfview source ↗
-
• twitter 1w positive@JasperTheVC @witcheer @NousResearch err wait yes you can use free models, with the paid cloud instances yes @JasperTheVC @witcheer @NousResearch err wait yes you can use free models, with the paid cloud instances yesview source ↗
-
• twitter 1w positiveWorking on AI research, post-training, or deployment? Come meet the @AfterQuery Team during SF Tech Week. Wed 10/7: Frontier of Finance with Goldman Sachs Fri 10/9: Rooftop happy hour We'd love to see you there. Details + RSVPs below. Working on AI research, post-training, or deployment? Come meet the @AfterQuery Team during SF Tech Week. Wed 10/7: Frontier of Finance with Goldman Sachs Fri 10/9: Rooftop happy hour We'd love to see you there. Details + RSVPs below.view source ↗
-
• twitter 1w positive❓ Question: Are you fully reading plans created by your agents? I love HTML plans and why we built it into Compound Engineering as an option. However, I admit I don't fully read most plans because, by the time the agent creates them, I've already settled so many questions in the ideation or brainstorming phase. Wha… ❓ Question: Are you fully reading plans created by your agents? I love HTML plans and why we built it into Compound Engineering as an option. However, I admit I don't fully read most plans because, by the time the agent creates them, I've already settled so many questions in the ideation or brainstorming phase. What I do read is the top part of the plans: what's being solved, the overall approach, etc. But the plans are so implementation-plan-heavy; that's more for the agents.view source ↗
-
• twitter 1w positiveUsing @pangram on LinkedIn is like seeing the AI matrix https://t.co/14PsIJNssi Using @pangram on LinkedIn is like seeing the AI matrix https://t.co/14PsIJNssiview source ↗
-
• twitter 1w positiveYour repeat button deserves a raise 🔁 Who's had your headphones working overtime this week? https://t.co/1jfHVB1vuN Your repeat button deserves a raise 🔁 Who's had your headphones working overtime this week? https://t.co/1jfHVB1vuNview source ↗
-
• twitter 1w positive@ngxson I love it but please have the model enumeration API return model type. We need to do silly things in pi to tell LLMs from classifiers. @ngxson I love it but please have the model enumeration API return model type. We need to do silly things in pi to tell LLMs from classifiers.view source ↗
-
• twitter 1w positive>be Jared >28yo in China >worked as AI engineer in big tech >quits the 996 and goes freelance >by day: does AI work in an illegal rooftop shack >by night: delivers food to pay for rent >cools his room with a water sprayer >but builds ESP32 cooling for his PC >the PC has better AC than he does >eats $0.70 boxed meals >A… >be Jared >28yo in China >worked as AI engineer in big tech >quits the 996 and goes freelance >by day: does AI work in an illegal rooftop shack >by night: delivers food to pay for rent >cools his room with a water sprayer >but builds ESP32 cooling for his PC >the PC has better AC than he does >eats $0.70 boxed meals >AI makes code cheap so his rates keep dropping >"it literally takes a single prompt" >rides 6-10pm >$15 a night >only delivers enough to pay for rent >"there's no going back" bro is delivering for to pay for rent so the coding job can stay.view source ↗
-
• twitter 1w positivedesign prompt of the day: ever noticed LLMs write text that's like: "More reliable. Less expensive." "Sleep better. Live longer." these are called stacatto sentences. tell your agent to 'never use stacatto sentences in landing pages' design prompt of the day: ever noticed LLMs write text that's like: "More reliable. Less expensive." "Sleep better. Live longer." these are called stacatto sentences. tell your agent to 'never use stacatto sentences in landing pages'view source ↗
-
• twitter 1w positiveI have added my own rules and skills to a new repo and already the AI coder is functioning better. First we classify the change (different tiers require different levels of planning and documentation). Then we request approval for the scope and a light plan. Then code is written or changed. Finally, a thermonuclear… I have added my own rules and skills to a new repo and already the AI coder is functioning better. First we classify the change (different tiers require different levels of planning and documentation). Then we request approval for the scope and a light plan. Then code is written or changed. Finally, a thermonuclear review and P0-P2s are addressed, with a PR and description.view source ↗