•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 701–725 of 1145 mentions
-
• twitter 1w negativeWeek 3 of Meta Ads Posting The recent ad got 2,000+ likes & 500+ clicks. Still, no one filled the form or messaged. 😭 And when I checked the people who liked the ad, they were mostly unrelated to our niche. META pushed the ad to the specified age group and location, but missed the most important thing "interests". Week 3 of Meta Ads Posting The recent ad got 2,000+ likes & 500+ clicks. Still, no one filled the form or messaged. 😭 And when I checked the people who liked the ad, they were mostly unrelated to our niche. META pushed the ad to the specified age group and location, but missed the most important thing "interests".view source ↗
-
• twitter 1w negativeStripe can shut down your business in one email and then refuse to tell you what you actually did wrong. That’s exactly what happened to us. MacWall had: • thousands successful payments • ONE $3.99 chargeback • real software + real customers Then @Stripe suddenly closed our account for vague “business model risk.” … Stripe can shut down your business in one email and then refuse to tell you what you actually did wrong. That’s exactly what happened to us. MacWall had: • thousands successful payments • ONE $3.99 chargeback • real software + real customers Then @Stripe suddenly closed our account for vague “business model risk.” Since then: • 5 days waiting • 10+ emails • multiple support chats • full compliance + policy cleanup • documents offered • formal complaint submitted And still no specific violation. No actionable explanation. No real answer. Meanwhile payouts are frozen and legitimate customers are being refunded i am not asking Stripe to expose some secret fraud algorithm. Just tell us what MacWall actually did wrong. If we violated a policy, name the policy. If something needs fixing, tell us what to fix. If the risk system made a mistake, have a human review it. A payments company this important should not be able to destroy a small business relationship with a vague sentence and then disappear. @stripesupport @Stripe we need an actual answer. Indie founders deserve better than this.view source ↗
-
• twitter 1w negativeTHIS IS OUR WHY. Sheridan Gorman was failed by open-border policies and sanctuary politicians who RELEASED this illegal alien TWICE before he carried out this heinous murder. We will never give up our mission to protect our nation and prevent tragedies like this from happening to more innocent Americans. THIS IS OUR WHY. Sheridan Gorman was failed by open-border policies and sanctuary politicians who RELEASED this illegal alien TWICE before he carried out this heinous murder. We will never give up our mission to protect our nation and prevent tragedies like this from happening to more innocent Americans.view source ↗
-
• twitter 1w negative2 years ago I quit my job, and only 2 months ago I finally removed the “Open to Work” badge on LinkedIn. It took me a long time to accept that I’m not an employee anymore. What helped was finally having stable income and realizing that, if needed, I could always find additional income sources. And worst case, I can al… 2 years ago I quit my job, and only 2 months ago I finally removed the “Open to Work” badge on LinkedIn. It took me a long time to accept that I’m not an employee anymore. What helped was finally having stable income and realizing that, if needed, I could always find additional income sources. And worst case, I can always go back to employment. I wouldn’t recommend quitting your job to build your own product. It’s probably the worst decision you can make. Unless you’re under 22. It took me 2 years to launch another product that actually makes money. And even then, I got lucky. And it was only possible because I had a stable income from my first app. I’d rather physically suffer from extra work in the mornings and evenings than mentally suffer from having no job, no income, and no success with my own business.view source ↗
-
• twitter 1w negativeEvery pattern in your life keeps repeating until you learn the lesson. Same type of person. Same money problem. Same argument with a different face. The loop doesn’t break when you understand it. It breaks when you choose differently. Every pattern in your life keeps repeating until you learn the lesson. Same type of person. Same money problem. Same argument with a different face. The loop doesn’t break when you understand it. It breaks when you choose differently.view source ↗
-
• twitter 1w negativeToday, Nolla Health became the first organization in the U.S. (and possibly the world) to receive regulatory approval for an AI to issue initial prescriptions. This makes Nolla the first ever actual end-to-end AI doctor. https://t.co/xn6rvb8IyW Today, Nolla Health became the first organization in the U.S. (and possibly the world) to receive regulatory approval for an AI to issue initial prescriptions. This makes Nolla the first ever actual end-to-end AI doctor. https://t.co/xn6rvb8IyWview source ↗
-
• twitter 1w negativeI'm confused why people pay so much for Kit and beehiiv. They're fine products. At 10,000 subscribers: ∙Kit Creator: $139/mo ∙beehiiv Lite: $139/mo Actually sending the emails: about $4/mo SendFox: $39/mo What am I missing? (Yes, I help run SendFox. Still asking.) https://t.co/KBmRyOlXTV I'm confused why people pay so much for Kit and beehiiv. They're fine products. At 10,000 subscribers: ∙Kit Creator: $139/mo ∙beehiiv Lite: $139/mo Actually sending the emails: about $4/mo SendFox: $39/mo What am I missing? (Yes, I help run SendFox. Still asking.) https://t.co/KBmRyOlXTVview source ↗
-
• twitter 1w negativeBeing right is not always worth the energy it takes to prove it. Being right is not always worth the energy it takes to prove it.view source ↗
-
• twitter 1w negativeopus 5.5 has a weird tradeoff nobody seems to talk about at max effort, it can burn through ~119k output tokens on a task or at xhigh is ~66k thing i’m wondering is how often those extra 50k+ tokens actually improve the answer, versus model just spending longer on the same problem opus 5.5 has a weird tradeoff nobody seems to talk about at max effort, it can burn through ~119k output tokens on a task or at xhigh is ~66k thing i’m wondering is how often those extra 50k+ tokens actually improve the answer, versus model just spending longer on the same problemview source ↗
-
• twitter 1w negativeso they just solved the problem of AI agents managing social media! Zernio lets Claude run your social media workflow, from creating content to actually posting it. basically, your AI agent can now build your product and run its social media too. https://t.co/NU0FdwVQ1B so they just solved the problem of AI agents managing social media! Zernio lets Claude run your social media workflow, from creating content to actually posting it. basically, your AI agent can now build your product and run its social media too. https://t.co/NU0FdwVQ1Bview source ↗
-
• twitter 1w negativeif the entirety of your engineering setup doesn't look like this you're falling behind https://t.co/6ZxUvtRrq2 if the entirety of your engineering setup doesn't look like this you're falling behind https://t.co/6ZxUvtRrq2view source ↗
-
• twitter 1w negativeCost is a bad thing It should be down oh no... 💀💸 https://t.co/xIbzD1vLEQ Cost is a bad thing It should be down oh no... 💀💸 https://t.co/xIbzD1vLEQview source ↗
-
• twitter 1w negativeI was recently building a coding agent and hit a problem that shows up fast: Long sessions make full-history context expensive and noisy. A common approach is to resend the entire chat history every turn. By turn 80, you're processing a lot of context just to recover the few details that actually matter. Oracle's Agent… I was recently building a coding agent and hit a problem that shows up fast: Long sessions make full-history context expensive and noisy. A common approach is to resend the entire chat history every turn. By turn 80, you're processing a lot of context just to recover the few details that actually matter. Oracle's Agent Memory takes a different approach: → Working: what the agent is doing now → Semantic: durable facts about the user/project → Episodic: what happened in past sessions → Procedural: rules for how the agent should behave Old turns can be summarized, durable facts extracted, and only relevant memory pulled back into the prompt. Oracle's 80-turn benchmark: ~1.3K input tokens/request with memory versus ~13.9K with full history. Better answers on 48 turns vs 13 in the benchmark. The bigger takeaway for me: agent memory is a system-design problem, not simply a bigger-context problem. Full breakdown + code. Checkout the blog here: https://t.co/lhU8RLnr4m In partnership with @OracleDevsview source ↗
-
• twitter 1w negativeI went from $1k to $15k MRR without adding a single new marketing channel. I just fixed my funnel. here's how: track 4 numbers every week: visitor → signup signup → trial trial → paid churn find the worst one. fix only that. then move to the next. each number tells you what's broken: visitor → signup low? your … I went from $1k to $15k MRR without adding a single new marketing channel. I just fixed my funnel. here's how: track 4 numbers every week: visitor → signup signup → trial trial → paid churn find the worst one. fix only that. then move to the next. each number tells you what's broken: visitor → signup low? your landing page isn't clear. signup → trial low? your onboarding is in the way. trial → paid low? they didn't see the value in time. churn high? your offer isn't worth staying for. if your funnel leaks, more traffic just means more people leaving. fix the bucket. then fill it.view source ↗
-
• twitter 1w negativeOpus 5.5 built this rotating model footer for a product. The pill swaps Gemini, DeepSeek, GPT and Grok without looking like a template. Prompt + the part I had to fix by hand in the reply. https://t.co/qbAmhz1y36 Opus 5.5 built this rotating model footer for a product. The pill swaps Gemini, DeepSeek, GPT and Grok without looking like a template. Prompt + the part I had to fix by hand in the reply. https://t.co/qbAmhz1y36view source ↗
-
• twitter 1w negativeRT @megadevhq: "Agents are incredibly talented engineers experiencing the world's worst Groundhog Day" 🎙️ @theo @joelhooks https://t.co/I… RT @megadevhq: "Agents are incredibly talented engineers experiencing the world's worst Groundhog Day" 🎙️ @theo @joelhooks https://t.co/I…view source ↗
-
• twitter 1w negative"We're watching and monitoring it closely." Secretary of State Marco Rubio confirms U.S. officials are closely tracking reports of a suspected plague breach at a Russian research laboratory. The concern follows reports that a Russian lab technician was infected and died, triggering localized quarantine measures in th… "We're watching and monitoring it closely." Secretary of State Marco Rubio confirms U.S. officials are closely tracking reports of a suspected plague breach at a Russian research laboratory. The concern follows reports that a Russian lab technician was infected and died, triggering localized quarantine measures in the region. U.S. health and security agencies are actively monitoring the potential risk of containment failure. "I don't think it's cause for alarm, but it is cause for focus and a cause to just keep an eye on. And we're doing that."view source ↗
-
• twitter 1w negativeimagine if tenso’s lists were an elaborate phishing scam this whole time and he was just playing the long game 🤔 imagine if tenso’s lists were an elaborate phishing scam this whole time and he was just playing the long game 🤔view source ↗
-
• twitter 1w negativeMarvel Comics and Webtoon are teaming up to release an all-new Marvel Comics App! Launching November 16, the app will feature more than 35,000 digital comics, personalized discovery tools, Pull List features, and new issues as early as 30 days after print release. Learn more: https://t.co/Q4DqMmDzJN Marvel Comics and Webtoon are teaming up to release an all-new Marvel Comics App! Launching November 16, the app will feature more than 35,000 digital comics, personalized discovery tools, Pull List features, and new issues as early as 30 days after print release. Learn more: https://t.co/Q4DqMmDzJNview source ↗
-
• twitter 1w negativeAI assistants completely change how notifications work. I've been using "do not disturb mode" for over 10 years now. I fucking hate notifications. Most of the apps I have are completely muted. The worst are emails, twitter or credit cards. I call those slot machine notifications. where 1/5 are actually important and … AI assistants completely change how notifications work. I've been using "do not disturb mode" for over 10 years now. I fucking hate notifications. Most of the apps I have are completely muted. The worst are emails, twitter or credit cards. I call those slot machine notifications. where 1/5 are actually important and the rest is slop. you can't completely mute them because you would miss the actual important stuff. The assistants completely change that. They learn what is important to you and what's not. A unified interface to your online life. my current set up is this. I got a dot with access to everything. all my messages, emails, linkedin etc. and computer use on my mac of course. same works with all the other assistants. no "do not disturb mode" anymore. if it pings me its important for me in that moment. "ping me 15 minutes before my meeting and brief me". "if its my family or my girl immediately ping me". "some weird transaction show it to me". for the other stuff it can brief me like twice a day. the biggest problem with "do not disturb mode" is that you live in constant "casino mode". what if there is something important I missed? you swipe up just to check. you are like a fucking skinner pigeon. Apple is an insane spot to implement this natively. It's obviously the next step. everything goes through your assistant. I've been trying it out this weekend. the results are great! my screen time and "number of times I check my phone" dropped a ton. I feel a lot more in the moment in general. get your life back! get out of the casino!view source ↗
-
• twitter 1w negativeGrok Bot can learn from your whole team One bad memory can mislead everyone What to save, what to keep private, and when to check the source A practical guide to Grok Bot memory engineering Bookmark and read below ↓ https://t.co/3SD7Ey9Q9N Grok Bot can learn from your whole team One bad memory can mislead everyone What to save, what to keep private, and when to check the source A practical guide to Grok Bot memory engineering Bookmark and read below ↓ https://t.co/3SD7Ey9Q9Nview source ↗
-
• twitter 1w negativethis is pure f*cking insane Anthropic just showed what 100 sub-agents running simultaneously inside Claude Code actually looks like an autonomous terminal spider crawls across 100 sandboxes in real time: ripping out fatal bugs, verifying AST diffs, and crowning the sole survivor run an elimination tournament with /a… this is pure f*cking insane Anthropic just showed what 100 sub-agents running simultaneously inside Claude Code actually looks like an autonomous terminal spider crawls across 100 sandboxes in real time: ripping out fatal bugs, verifying AST diffs, and crowning the sole survivor run an elimination tournament with /arena claude /arena --agents 100 --seed 7 --waves 10 Opus 5.5 spawns 100 diverse problem-solvers across 2,160 strategy decks Sonnet 5.5 pairs them in head-to-head duels inside isolated sandboxes Adversarial judges attack every solution at three critical choke points: > before code executes: does the architecture hide silent race conditions or lock contention? > when input scale multiplies: does memory explode to O(N^2) or leak file descriptors? > before marking done: does a single unhandled timeout crash the entire pipeline? 100 spawn. 50 attack. 1 battle-tested survivor ships The core engineering shift: zero context bleeding If 100 agents share one chat window, they copy each other's hallucinations within 2 prompts Arena locks each agent into an isolated sandbox, forces 5 rounds of brutal peer critique, and lets a blind judge eliminate flaws Spawn on diversity. Eliminate on fatal flaws. Ship the survivor - the full tournament loop > wave 1-10: 100 diverse briefs generated across 2,160 reasoning seeds > round 1: 50 head-to-head matches paired with segregated reasoning styles > round 2: autonomous spider crawler inspects AST trees and extracts fatal bugs > round 3: defenders get one revision round to patch verified vulnerabilities > round 4: blind judge scores survivors on a strict 100-point rubric > grand finals: 1 hardened, attack-proof solution survives into production paste the setup and this prompt into Claude Code below: "Configure my Claude Code workspace for the 100-agent tournament arena: 1. Audit ~/.claude/skills and project .claude/skills for the arena-skill package > Clone from repo if missing > Configure 10 execution waves with 10view source ↗
-
• twitter 1w negativeRetry on 429. Server is already shedding load. Client retries immediately. Retries make it worse. Then the client retries the retries. 429 means slow down, not try harder. Need: • Retry-After • jittered backoff • a cap • no retry on 4xx that isn't 429 or 408 Have you taken down your own API with a helpful ret… Retry on 429. Server is already shedding load. Client retries immediately. Retries make it worse. Then the client retries the retries. 429 means slow down, not try harder. Need: • Retry-After • jittered backoff • a cap • no retry on 4xx that isn't 429 or 408 Have you taken down your own API with a helpful retry?view source ↗
-
• twitter 1w negativeyour agents don't have a memory problem. they have an ownership problem. every agent you add starts as a stranger, because your context lives inside its app Fulcra moves it out: one context store that belongs to you, not your apps > every agent, same context: Claude, ChatGPT and Grok over MCP, Hermes and OpenClaw o… your agents don't have a memory problem. they have an ownership problem. every agent you add starts as a stranger, because your context lives inside its app Fulcra moves it out: one context store that belongs to you, not your apps > every agent, same context: Claude, ChatGPT and Grok over MCP, Hermes and OpenClaw over the CLI > teach once: health, calendar, location in one store, every new agent skips onboarding > your keys: choose what's shared, keep access scoped, revoke any time > agents that work together: one team on one project, messages to someone else's agent, groups across accounts > builders get the Life API: real-time data on every consenting user start free with 5 GB. Pro is $14.99/month the video walks through all of it in 59 seconds stop re-introducing yourself.view source ↗
-
• twitter 1w negativeEver had a prompt you knew was bad but were scared to tweak? Yep, that's me like 90% of the time. And apparently, I'm not alone... I recently spoke with @Al_Grigor about the RAG app he built for @DataTalksClub. His first version worked fine. Then corrections started piling up. So he added another instruction to … Ever had a prompt you knew was bad but were scared to tweak? Yep, that's me like 90% of the time. And apparently, I'm not alone... I recently spoke with @Al_Grigor about the RAG app he built for @DataTalksClub. His first version worked fine. Then corrections started piling up. So he added another instruction to the system prompt. Then another. Then another... Until the prompt became a giant monolith he knew needed cleaning up. But the system worked... Which left him with the classic engineering problem: “When I change it, what if something else breaks?” Without evals, he had no way of knowing. Or, as @Al_Grigor put it, “Without evals you are blind.” But there was another problem... He hadn't recorded the traces he needed to build an eval dataset. So Alexey used what he already had: • GitHub issues → inputs • Accepted PRs → successful outputs • Manual corrections → failures • Human edits → expected outputs This gave him enough proxy data to bootstrap his eval set. Today, he uses Opik by @Cometml to trace his RAG workflows, version his eval datasets, and compare experiments. And whenever something breaks, he follows a simple process: 1. Capture the failure as an eval case 2. Fix it 3. Run the individual case 4. Run the entire eval set 5. Make sure nothing else broke @Al_Grigor calls this “eval-driven development.” I love that framing. Because evals don't just tell you whether your AI system works... They give you the confidence to change it without wondering what else you broke. In tomorrow's Decoding AI Magazine, @Al_Grigor and I break down his entire RAG eval workflow, including how he built an eval dataset when he had zero traces. Subscribe here to get it in your inbox: https://t.co/S0eIb9cEoCview source ↗