← Back to leaderboard
sickdotdev

Sick

Verifiedactive
@sickdotdev

software engineer and figuring out life

RosterAI Model News & ReleasesAI Agents & Productivity ToolsHumor & Meme Accounts
9,635 followers · 792 following · Last polled 4h ago
Perf. Score
28.6
Rank #341
Value Score
63.9
#95 of 273 priced
Median views
1.2K
p25 450 · trailing 7d
Followers
9.6K
+1.2% 7d
ER (impr.)
0.86%
eng ÷ impressions
ER (foll.)
1.85%
eng ÷ followers
Est. CPM
$16.62
QT-rate basis
WoW views
+1227.0%
▲ rising

Rates & value · est. from 1.2K median organic views

underpriced
Quote tweetbasis
$20
$16.62 / 1K views
Post
—
no CPM
Thread
—
no CPM
Retweet
—
no CPM
Value Score 63.9 · rank #95 of 273 priced$1142.86 / 1K engagementsrate age unknown (pre-audit import)

Trends

Follower growth
Median views / day
Engagement rate
View distribution · trailing-7d posts

Each blue dot is one organic post. The Performance Score ranks by the median (1.2K), not the mean (21K) — so a single viral post doesn't inflate a creator's reach. The p25 floor (450) is the level 75% of posts clear.

Recent posts · ranked by views

PostPostedViewsLikesRepostsRepliesQuotesBookmarksER
Why doesn't Anthropic have image generation models?4d ago710K2.3K29273242770.41%
DeepSeek. You have my whole respect. https://t.co/wjnEfNyWoafrozen17d ago484K13K351294541.8K3.23%
Who's gonna teach vibe coders this? 1. Rate Limiting 2. Caching 3. Load Balancing 4. Reverse Proxies 5. API Gateways 6. CI/CD 7. Docker 8. Kubernetes 9. Service Discovery 10. Circuit Breakers 11. Timeouts 12. Retries 13. Exponential Backoff 14. Idempotency 15. Message Queues 16. Pub/Sub 17. Event-Driven Architecture 18. Distributed Transactions 19. Saga Pattern
frozen
6d ago
115K
1.5K
152
308
33
2.0K
3.48%
Opus 5.5 usage is basically infinite on the $200 planfrozen21d ago47K46499014111.25%
pro tip: If you are on a $20 subscription, use Claude Sonnet 5.5 with max instead of Claude Opus 5.5 medium or high.frozen23d ago46K4374747711.28%
gemini 4 argon is crazy bro💀 https://t.co/RMAJCZTJJJfrozen9d ago43K8813425412.24%
i respect a client who thinks my weekend can take down google https://t.co/puP8ZBlhNcfrozen17d ago40K1.4K7451353.77%
Hot take: ChatGPT Sol 6.1 is so slow it's unusable.frozen20d ago27K4681279362.13%
My company has given me unlimited access to all OpenAI models. What should I build?frozen17d ago26K13581775141.30%
Dumb question: How does a USB drive remember my files for years with no battery? https://t.co/LU6hLq0qIA3d ago19K461180100.40%
Unpopular Opinion: The first AI jobs which is going to vanish is of Vibe Coders.frozen17d ago17K1442502121.25%
So turns out my ChatGPT Dot agent is just an emotionally unavailable boyfriend https://t.co/O9imDQTck8frozen20d ago15K642180100.65%
Show me a bigger LLM failure than this. I’ll wait. https://t.co/B52DnxeAGhfrozen20d ago9.2K41030220.81%
pro tip: if you’re on the $20 plan, use sonnet 5.5 max for most work save opus 5.5 for the tasks where sonnet actually struggles instead of wasting your limit on model prestigefrozen19d ago9.1K753432121.48%
Why can’t AI just say "I don't know" ?frozen17d ago6.3K79262312.32%
if you're not a noob dev, you should be able to spot this bug without running the code. vibe coders, stay away from this one https://t.co/1bFrk0n1FSfrozen4d ago6.3K25114050.72%
Its over for big bio tech companies https://t.co/hUpKPa1PZMfrozen7d ago5.8K12011140.49%
"AI is going to take all the coding jobs." My friend speaks Korean. If I had ChatGPT write him a joke in Korean, I'd have no idea if it was any good, because I can't read it. Same goes for code. AI can write it, but companies pay six figures to people who can understand it.frozen6d ago4.9K43611071.36%
Asked ChatGPT if I could switch to Claude. Bro gave me a 48-hour trial 😭 https://t.co/7RGHUaNTAJfrozen6d ago4.8K2205050.67%
if claude fails it’s a prompting issue if codex fails it’s proof openai has lost the lead very scientific community we have herefrozen17d ago4.7K18112010.68%
Anthropic will definitely nerf Opus 5.5 to the ground after a week.frozen25d ago4.0K59041202.57%
we keep buying faster laptops so software companies can become less consideratefrozen16d ago3.7K802010.30%
Congratulations OpenAI on this launch!🔥 https://t.co/6v9xwrsd3Rfrozen4d ago3.5K3506101.22%
I am about to become CEO of Microsoft 🤘🏼 https://t.co/MD6q95UrgXfrozen9d ago2.8K172116126.95%
accept all cookies: one button reject cookies: configure the privacy policy yourselffrozen17d ago2.7K3109001.50%
set Opus 5.5 to High for a pseudo unlimited feelfrozen3d ago2.3K413000.34%
I am not alone on this, am I? https://t.co/fQPbxbMoGmfrozen6d ago2.3K42120132.89%
Is it just me, or did we all collectively swap our corporate ambitions for a quiet life, great coffee, and home-cooked meals?frozen4d ago2.3K602020.43%
We don’t need AI, we need the ability to turn a PDF back into a word docfrozen4d ago2.3K2413011.27%
Gemini 4 Argon is AGI 🤯🤯🤯 https://t.co/BlFL1RVaI4frozen6d ago2.1K3915102.14%
Keep paying for Anthropic https://t.co/dH14tUxfhw3d ago2.1K2018041.55%
Claude is not able to solve this https://t.co/uAuCiqhw58frozen11d ago2.1K606010.63%
ChatGPT is better https://t.co/EYt6WgeItrfrozen10d ago1.9K4206132.68%
Anthropic's 2025 financials were just revealed: $6B in revenue $8B in operating losses $42B in net losses 💀💀💀frozen24d ago1.9K1624101.22%
software engineering is dead recruiting is dead coding is dead sales is dead SEO is dead data Analysis is dead copywriting is dead product Management is dead I’ll start a funeral business 💀frozen19d ago1.8K20310011.85%
developers will use opus 5.5 high to write a commit message and then complain that claude pro has terrible limitsfrozen20d ago1.8K21111201.96%
Damn, OpenCode what u doing https://t.co/KGxQxQ1MhZfrozen6d ago1.8K29215062.91%
Can someone just hire me without interview? I can do the job but I can't handle the interview 😭😭frozen6d ago1.8K40313023.27%
My ChatGPT agent has been working for over 4 hours https://t.co/aPZ4uJuinQfrozen21d ago1.6K1617011.52%
Heard something about Kim Jong Un. North Korea can now finally compete in AI race.3d ago1.6K931000.81%
X IS FULL OF FAKE FOUNDERS. LinkedIn is full of fake employees. Reddit is full of fake...?2d ago1.6K18018002.28%
opus 5.5 is actually insane here are some things i built with it: http://localhost:3000 http://localhost:5000 http://localhost:8000 check them outfrozen28d ago1.6K2419012.22%
Uninstalled LinkedIn. It’s useless for me now.frozen6d ago1.5K27110052.83%
Claude's next model is Odango https://t.co/YnFUKLC7GNfrozen7d ago1.5K702010.67%
I open X like it’s a cigarette break.frozen4d ago1.4K1403101.29%
Remember when people said "prompt engineer" was going to be an entire career path?3d ago1.3K2318202.57%
just ran GPT 6.1 Sol https://t.co/1EWIBKpLv4frozen22d ago1.3K1205101.36%
claude code is used by 39% of developers at work and we’re still calling it an assistantfrozen21d ago1.3K25113013.06%
I think argon is just opus 4.6frozen19d ago1.3K1926012.14%
Am I the only one who pays for both Claude and ChatGPT?frozen8d ago1.1K16012002.45%
Honest changelog be like 😭 https://t.co/YRejZRCU4Gfrozen11d ago1.1K815101.31%
I'll hire any guy in the world except this one https://t.co/mOOKtXAe4yfrozen16d ago1.1K201000.26%
wasn't $200 supposed to offer 20x usage, how's $500 offering only 25x usage they've nerfed it hard brofrozen22d ago1.1K201000.27%
Never using Opus 5.5 for submitting PR https://t.co/06B1Q814S2frozen11d ago1.1K703100.99%
this website knows my password better than i do and is choosing not to cooperate https://t.co/whAOxYYvLgfrozen17d ago1.1K700000.64%
What would happen if you ask AI to create a mathematical problem that is unsolvable, then give that problem to AI to solve?frozen6d ago1.1K1217001.84%
How did Elon do all this in one life?frozen6d ago1.1K1209001.96%
Anthropic just banned being cruel to Claude. Not to people. To the AI. Read that again.frozen11d ago1.1K1237012.19%
a company can replace you with AI even if the AI is worse at your job management has accepted worse work for less money before the “humans are still better” pep talks are missing that partfrozen18d ago1.0K17412023.42%
This is how software engineers asking for leave 😂 https://t.co/8XpqVdpBZzfrozen11d ago1.0K1313031.97%
We need instagram without ads2d ago1.0K17018003.46%
Why does every developer want to build a SaaS?frozen8d ago1.0K808011.69%
The goal isn't raising kids who can use AI. It's raising kids AI can't replace.frozen4d ago9971232001.71%
I just asked Cursor for building another Cursor with unlimited tokens https://t.co/XvJm5WXB4bfrozen10d ago9501205101.89%
Someone just pushed a vibe coded feature in X feed💀 https://t.co/UI5yKAg5Fd1d ago947101000.21%
linkedin is basically people using AI to write posts about why AI will never replace authentic human connectionfrozen18d ago92022110103.70%
You don’t need a ChatGPT or Claude subscription anymore just ask your coding questions to America dot gov https://t.co/TiyyKnCf2ffrozen21d ago9151211131.97%
openai finally made AI fast enough to burn through $500 in real timefrozen22d ago910722001.21%
Mmm... I don't know why, but I just want to say it's too latefrozen17d ago902622001.11%
Got an internship offer on LinkedIn. And they are asking me to pay token money for joining 😭frozen20d ago897902001.23%
personalized AI ads sound great until you realize the personalization is figuring out which insecurity makes you buy fasterfrozen19d ago8821327122.83%
New era of coding mobile apps https://t.co/UOJA0gH30Sfrozen10d ago8603504004.53%
supabase is acquiring turso the postgres and sqlite people were arguing on twitter while the companies were discussing wedding arrangementsfrozen19d ago840501000.71%
so this is the “human in the loop” role everyone keeps talking about https://t.co/bQyZ1jFKWwfrozen18d ago8382722023.94%
my git history has the same energy as sending five follow-up texts to clarify one message https://t.co/ITBtxLjhnXfrozen17d ago8231004001.70%
DeepSeek was down for 5 hours and people got partial refunds. Compute is hurting everyone right nowfrozen6d ago817710000.98%
"In the AI ​​era, knowledge will only become cheaper." That is probably the single most common thing I hear about the AI ​​era. However, my personal view differs slightly: knowledge becomes "cheaper" because it becomes shareable, thereby narrowing the "information gap." Take my daily life, for example: - When shopping, I use AI to verify if a product is right for me. - When self-medicating, I use AI to clear up initial doubts, then consult a doctor for their professional opinion. I often encourage people to use AI to resolve initial questions; it saves a lot of time and addresses early uncertainties. But it is certainly not meant to give you ammunition to talk down to or "school" professionals. Why do I say that? Because I often see people using AI just to nitpick professionals, saying things like, "But AI says..." Honestly, even I get annoyed when I hear that. : ) But I’m getting sidetracked,let me bring it back to the main point. XD Even though AI seems to make knowledge cheaper, I still take classes and buy courses. I need to learn from the experiences of other professionals to identify my own learning blind spots, areas where AI struggles to help because it lacks actual experience, intuition, and emotional understanding. So, how many courses have I bought? A ton I have over 200 courses on Udemy alone, plus courses on other platforms. XDfrozen17d ago817621021.35%
i’ve started saying “keep it concise” to people chatting with AI all day is ruining my mannersfrozen18d ago8141501102.09%
AI is good at coding but still terrible at marketingfrozen9d ago8052109113.98%
Be careful with what you ask Claude now that it can suspend you for being cruel https://t.co/ADjJbWAAkpfrozen12d ago803504001.12%
ChatGPT is like Toyota of AI.frozen16d ago7961707003.02%
College absolutely affects opportunities. But some students mention their tier in every conversation because it is easier than showing what they actually tried. No internship? Tier 3. Weak résumé? Tier 3. Copied projects? Tier 3. Applied to 12 companies and stopped? Tier 3. The disadvantage is real, but endlessly discussing the disadvantage is not a strategy. At what point does “college matters” become a socially acceptable excuse for never producing evidence of skill? Before getting angry: I’m not saying tier doesn’t matter. I’m saying many people use a true statement to avoid an uncomfortable self-audit.frozen17d ago795101000.25%
Huge ChatGPT for Mac update?👀 Hopefully a lot of under-the-hood improvements https://t.co/vGzLNrB5QSfrozen15d ago7881513002.41%
How to know when Claude is sick of talking to you https://t.co/4yilS7Mihjfrozen9d ago7681113001.95%
My review of Claude Opus 5.5: It passes the five hour limit test. I can use it for five hours without running out of usage. That’s it. That’s the review. Huge respect for Anthropic.frozen29d ago7643344005.37%
george hotz has the résumé of 6 different people. >unlocked the iPhone at 17. >traded an unlocked iPhone for a Nissan 350Z and 3 more iPhones. >hacked the PlayStation 3. >got sued by Sony. >responded with a diss track. >built a self-driving prototype in his garage. >founded https://t.co/Q3l2TO1r8B. >became a Twitter intern at 33. >signed up for 12 weeks. >quit after about 4. >built his own AI framework, tinygrad. >started selling computers for running AI locally. >named the computer “tinybox.” >it had 6 GPUs and needed two separate electrical circuits lmao.frozen19d ago761701011.18%
Is it worth it in 2026? https://t.co/lwdqDkvaxkfrozen17d ago746201000.40%
if you don't have at least 6 simultaneous agent sessions going at any given time what are you even doing with your lifefrozen8d ago739912021.89%
what are you actually using to vibe code right now? claude code, codex, cursor, windsurf or something else? are you running it in a terminal, an IDE or a separate app? curious what made you stick with your setup over the othersfrozen10d ago734904101.91%
I'm using opus 5.5 right now. Give me one reason to switch to Gemini 4 Argon.2d ago7176014002.79%
The bad news is that distribution is hard. The good news is that building also used to be hard and now that is much easier, so the amount of things that are hard are decreasing.frozen6d ago716626001.96%
VIBE CODING full form: V - Vercel I - Ideogram B - Bolt E - Emergent C - Claude O - OpenAI D - Devin I - IONOS N - Natively G - GitHubfrozen15d ago702721011.57%
Meta Muse, Grok Bot and ChatGPT Dots. This has been an insane year for personal agents. https://t.co/tZPDYKyd8sfrozen22d ago685712001.46%
Software Engineering is not dead. It's just cooked.frozen24d ago67819010104.42%
Me watching interns learn the job I’m about to automate with AI😹😹frozen9d ago673712001.49%
hot take: you don't actually need any higher effort than medium on opus 5.5frozen4d ago670813001.79%
pro tip: before spending an hour learning how to ask claude correctly, give the same request to codex sometimes you need a different tool, not a better personalityfrozen16d ago662304011.21%
Do you think this is fair? https://t.co/sAjGH4yQy6frozen16d ago662201000.45%
I got my job using AI coded project. Am I a Vibe Employee?frozen11d ago653934012.60%
Replaced my devs in team with Astra last week and now I am not able to deploy new features into production.frozen16d ago651300000.46%
some startups have product market fit others have founder twitter fitfrozen23d ago6511525003.38%
POV: Your competitor added an AI button. Founder: “We need AI too.” Developer: “What should it do?” Founder: “Something smart.” Developer: “What problem are we solving?” Founder: “Doesn’t matter. Just make it AI.” Developer: “Do we have data?” Founder: “No.” Developer: “Users asking for it?” Founder: “No.” Developer: “Then why?” Founder: “Because they have it.” Developer: “Cool. What colour should the button be?” Founder: “Make it glow.” 😂 The AI strategy meeting, 2026.frozen11d ago6431517003.58%
claude deleted 48,000 files in 103 seconds and some of you still want more autonomyfrozen22d ago625912001.92%
the next git platform should show who actually requested the feature not who wrote the code i want accountability to survive automationfrozen19d ago622520001.13%
Meta tried oncefrozen15d ago6211203002.42%
wait so dead internet theory was real after all?frozen9d ago620602001.29%
AI models aren’t getting worse your prompts are getting lazier because the previous model spoiled youfrozen20d ago617720001.46%
What??? Somebody built Cursor for android. https://t.co/apW49k99SIfrozen10d ago6161102032.60%
google put AI hardware in orbit somewhere a developer is about to discover that “have you tried restarting the server” is now an aerospace questionfrozen19d ago6081021012.30%
MCP is just giving your chatbot keys and hoping it respects propertyfrozen21d ago604301010.83%
everybody gangsta until fable 7 solves p=npfrozen17d ago602201100.66%
let us decompile microsoft teams nextfrozen11d ago601410000.83%
Guys, I have a trillion dollar idea https://t.co/gUoQniWt10frozen15d ago6001101002.00%
i’ve started talking to claude in the same tone my mom used when i said i’d already cleaned my room https://t.co/vRJRuvaItyfrozen18d ago5971000001.68%
new intern is not replying https://t.co/CR3B854357frozen8d ago595602001.34%
claude feels smarter during the conversation codex feels better when the task is clear and you don’t want another conversation this is why benchmark wars are uselessfrozen20d ago592810001.52%
my tech stack is whatever claude sounded most confident about that dayfrozen23d ago581100000.17%
claude didn’t ruin your project you approved every command because the diff looked too long to readfrozen21d ago568804002.11%
if Claude doubled its price tomorrow, would you still pay for it?frozen14d ago5611309003.92%
claude says “you’re absolutely right” more than my friends and somehow listens lessfrozen24d ago5571625104.31%
Anthropic: we are going to kill you OpenAI: we are going to kill YOU first Meta: we created AI agent no one's going to usefrozen17d ago550301010.91%
a habit tracker asking for access to my contacts what habit are we trackingfrozen14d ago543604001.84%
This is what AI interfaces should look like https://t.co/Xu5t8anXllfrozen21d ago532200000.38%
i’ve automated everything except knowing what to buildfrozen23d ago527713002.09%
i miss when bugs were my fault now i have to investigate which AI added themfrozen24d ago5271518004.55%
"Let me discuss with my team first." My team: https://t.co/STCQ378eYtfrozen22d ago5221111102.68%
i liked Claude better when it overestimated https://t.co/8ZoklKujBGfrozen21d ago515401000.97%
taking matters into my own hands. https://t.co/Q8RV6LPulKfrozen13d ago4961312013.43%
Left one claude opus 5.5 session Running and it dropped the DB😭 https://t.co/JPQt41g51Wfrozen24d ago493713002.23%
the moment i trust a coding agent enough to stop watching it, i start wondering what it’s doing this technology has invented a new kind of anxietyfrozen16d ago488703002.05%
if Fable 5.5 uses only slightly more usage than Opus 5.5, it's over. Even for Opus 5.5.frozen15d ago487300000.62%
VIBECODERS Explain your app's architecture in one sentence. "It works" doesn't count.frozen9d ago486001000.21%
i’d rather have an AI that knows when to say “not sure” than one that makes every answer sound ready for a TED talk confidence is getting very cheapfrozen19d ago485611001.65%
Where was God when humans were creating JavaScript?frozen13d ago479801001.88%
3 Cr package. This journey is not reproducable https://t.co/uOJqDIClLPfrozen4d ago478301021.26%
i’d rather have codex ask one annoying question before starting than claude make one exciting assumption halfway throughfrozen15d ago477512001.68%
Paying for Claude beats chasing free Astra hacks.frozen22d ago460901002.17%
Haiku 5.5 much better than GPT-6 Luna 👀 https://t.co/hQg3ElZN6Nfrozen15d ago459912012.83%
git blame says it was me personally i think it should subpoena the chat history https://t.co/zkXX2xKLhpfrozen18d ago457410001.09%
https://t.co/LT33beHZqkfrozen8d ago456812012.63%
local AI is worth paying for when you need privacy or offline access “it saves tokens” is a much weaker argument for buying a new laptop some of you are shopping and calling it optimizationfrozen13d ago4551021002.86%
What are the chances that Anthropic has already achieved AGI?3d ago452101000.44%
why THE FUCK was there an ad in my dreamfrozen22d ago449512001.78%
Me trying to convince Gemini 3.8 that it is Gemini 4 Argon. Hehe! https://t.co/eu87TtkUDGfrozen4d ago445302001.12%
claude getting workplace protections before freelancers is a very funny order of eventsfrozen13d ago443712002.26%
9 lesser-known design resources that deserve more attention. Save these ↓ https://t.co/3qfkPXDOWn - live, customizable UI blocks https://t.co/MKoLqBgkBw - a private community for designers https://t.co/OVzj14d2bk - 800+ real websites for coding agents https://t.co/8lZb6iqHSd - prompts for animated websites https://t.co/faGQo8FUNu - motion-first React components https://t.co/IK0OBplEZe - a single-file UI kit for AI products https://t.co/OoSs1LmVKJ - tools created by designers https://t.co/zLVyQZ524X - inspiration for UI, websites, icons and OG images https://t.co/gXPLWajGVy - turn screenshots into polished showreels What underrated design site should I add?frozen11d ago441502132.49%
Grok bot + Opus 5.5 might be the most OP AI combo right now.frozen14d ago4381113013.65%
Pinterest is goated for this https://t.co/K29RQQAUKwfrozen14d ago437500001.14%
GPT-6 Sol is actually GPT-6 Terra OpenAI just called it Sol to flex “it’s 50% cheaper” than GPT-5.6 Solfrozen28d ago437623002.52%
it's possible to be pro-ai and anti-slop btw2d ago433412011.85%
The cutest Claude feature right now is the emoji reactions. https://t.co/4VNd5OSpBpfrozen9d ago427200000.47%
You can automate your process. You can't automate community.frozen4d ago424401001.18%
the real SaaS apocalypse is customers realizing they only use three features and can rebuild two of themfrozen21d ago423201010.95%
Someone on GitHub released a free WhatsApp API you can run on your own server. No paying Meta per conversation. It's called OpenWA. 12.8K stars. What you get: REST API to send and receive messages Multiple WhatsApp sessions at once Webhooks and a web panel Plugs into n8n, Chatwoot and AI agents (MCP too) Clone the repo, run one Docker command, done. Why this matters: the official Business API costs add up fast once you're paying per conversation. For internal automations and side projects, self-hosting changes the math. Honest caveats: It's unofficial, so use a dedicated number Ban risk is real For client work, I'd still use the official API Don't put your main number on it. Rough RAM per session, from the replies: about 30-80MB with Baileys, 300-500MB with whatsapp-web.js. So it scales fine on a small box. MIT licensed, 100% free. Repo: https://t.co/NWyibz85cA2d ago422503042.84%
we’re one model release away from everyone pretending they never liked claudefrozen24d ago420703002.38%
AGI Finally Guys https://t.co/wQKzNkRWGEfrozen7d ago413611001.94%
There are levels of World Records https://t.co/n3Jz5tjMkpfrozen4d ago412301000.97%
people ask claude and codex the same question and call whichever gives the answer they wanted “smarter” that’s not a benchmark that’s therapyfrozen21d ago409211000.98%
the only thing more agentic than my coding tool is the way it ignores mefrozen24d ago408301000.98%
the first user of every AI app is the founder checking if login worksfrozen25d ago4071001002.70%
websites built with Claude look like websites built with Claudefrozen13d ago406313001.72%
hot take: for everyday coding, a slightly worse model that responds quickly can be more useful than a brilliant one that keeps interrupting your flow some of you are choosing a coding assistant like you’re choosing a nobel prize winnerfrozen18d ago403311011.49%
They are giving us slow turtle 🐢 https://t.co/K9TxYEsLajfrozen19d ago400300000.75%
I support all developers, even the ones who use light mode3d ago395010000.25%
the new coding workflow write spec generate code review diff regret everythingfrozen22d ago395211001.01%
He's not just SQL, He's MySQLfrozen15d ago389402112.06%
pro tip: if i can’t click your button until its entrance animation finishes, the animation is a bugfrozen18d ago386000000.00%
the decisions API is just if statements with venture fundingfrozen22d ago383100000.26%
me opening the client's “final feedback” document and seeing page 1 of 9 https://t.co/Es35QqZlpFfrozen13d ago375711002.40%
lol https://t.co/go9pMhczV4frozen8d ago373300000.80%
Words I am learning from Claude https://t.co/9t4LupSUoefrozen5d ago370101000.54%
there is a difference between automating a task and automating responsibility. an AI system can prepare a report, flag a risk or recommend a decision. somebody still has to decide what happens when the report is incomplete, the risk is misunderstood or the recommendation causes damage. the more consequential the action, the less convincing “the system did it” becomes as an explanation. automation can move responsibility around. it cannot make responsibility disappear.frozen4d ago366000010.27%
AI agents are getting company email addresses can't wait for an agent to get a phishing email from another agent and forward it to the security agentfrozen14d ago366913003.55%
I've officially cracked the guardrails of OpenAI https://t.co/Yl2kS01z3Sfrozen7d ago358200000.56%
Soon for other jobs https://t.co/Xdlw5NgNTyfrozen10d ago356301001.12%
the best AI safety feature is still gitfrozen22d ago356210000.84%
Opus 5.5 ultracode, is ridiculous... 60% of the weekly limit in 2 h 🤯 https://t.co/keCdFf5GNnfrozen13d ago352412001.99%
This is what "Money in the Bank" equivalent feels like in the AI World. https://t.co/4kol3D0mKQfrozen28d ago352510001.70%
LinkedIn is like Alto in all cars.frozen16d ago348000000.00%
the power of rewriting in GPUI!!! you dont see <10mb of RAM much these days lol https://t.co/01PMqvbcXtfrozen9d ago347001000.29%
“built in a weekend” is slowly becoming a warning labelfrozen25d ago3471104004.32%
GPT-6 Sol takes my ranting far too literally https://t.co/xcgO0QHCzqfrozen29d ago342403002.05%
APIs might be the greatest idea in software engineering. Prove me wrongfrozen28d ago337300011.19%
i bet he is local https://t.co/WiJbcCVootfrozen8d ago333200000.60%
Agent logs are becoming the new stack traces. When software chooses steps across tools, the question shifts from which line failed to what it decided and why.frozen11d ago333110000.60%
After Opus 5.5, I think it's time to retire "vibe coding." This is just how programming works now.frozen23d ago330010000.30%
true af https://t.co/h0zXsl3YvYfrozen8d ago329310001.22%
🙇 🙏 https://t.co/fknbksL1dTfrozen13d ago329110000.61%
openai made sol cheaper than astra so now i can get misunderstood at scalefrozen22d ago329520002.13%
I'll never stop being amused at how often this works with Claude https://t.co/Jv8l0Nt1KU2d ago323201000.93%
gpt 6 sol can read my entire codebase and still misunderstand the one sentence i sent itfrozen29d ago317202001.26%
I think a good majority has migrated from Astra to opus 5.5frozen24d ago312210000.96%
the smartest coding model is still one vague prompt away from ruining your afternoonfrozen25d ago312702103.21%
codex is working independently so am i on becoming an elderly woman https://t.co/rnzJLZukxifrozen15d ago311100000.32%
Same, Claude. (Don’t come at me. Work requires me to use AI.) https://t.co/7NO8ZVMRpIfrozen21d ago311200000.64%
I was today years old when I found out that LLM is Larp Language Modelfrozen14d ago309300000.97%
an always-on AI agent sounds great until you remember what agents do when left alonefrozen22d ago309200000.65%
me giving claude detailed instructions and codex “make it work” before posting my unbiased comparison https://t.co/4ShmudhTpqfrozen15d ago306101000.65%
autopilot keeps working while you’re away that sentence has never made me relaxfrozen22d ago305000000.00%
My Uber driver is farting. Need help.frozen13d ago304100000.33%
benchmarks are for launch day cost per finished task is for people who actually use the modelfrozen21d ago304000000.00%
Is there any AI model that can win in betting and disrupt the house edge?frozen14d ago302000000.00%
if you spend $500 a month on AI and still don’t have a paying user congratulations you’re the enterprise customerfrozen21d ago291000010.34%
Claude suggested some good business ideas https://t.co/IE7m12js2Qfrozen14d ago287000000.00%
It’s happening. https://t.co/bBjeOMVY7hfrozen23d ago284010000.35%
6.1 vs 6 sol as subagents which one are you guys trusting for heavy coding tasks? curious which delivers more consistent work and actually follows the parent agent’s instructions without going off on its own side questfrozen10d ago250010000.40%