Rulya

1K posts

Rulya banner
Rulya

Rulya

@Rulyaxd

never look back

Fomo Katılım Nisan 2022
771 Takip Edilen82 Takipçiler
Sabitlenmiş Tweet
Rulya
Rulya@Rulyaxd·
Four trades. +$117, -$229, -$130, -$114. Final caption: LOSS. The TikTok this comes from is titled "17 year old day trader retires parents using AI." The chart watermark reads MNQM25, the Micro Nasdaq June 2025 contract. The top of his screen reads BAL $98,869, MLL $97,000. Eighteen hundred dollars between him and a margin call. That spread is what makes the account a prop firm funded challenge. He is renting the $100,000, not earning interest on it. The challenge ends the moment balance touches the MLL line. The AI tool he runs is called Scalp Trading. The dashboard returns Pattern Analysis on a one-minute timeframe with a Sell signal, entry 21000, stop 21006, risk 0.25 percent of account. Recommended dollar risk: $25. 0.25 percent of $98,000 is $245. The tool is sizing his trade against a $10,000 account that exists only inside its risk engine. The number is two zeros short of his real balance. He doesn't catch it and takes the Sell signal. The captions roll. +$117, then -$229, -$130, -$114. The screen reads LOSS. The bedroom behind him still has the LED string lights and the tapestries. The Stop Loss bracket on the chart is doing the actual parental retirement. You can check this math without leaving the page. 0.25 percent of $100,000 is $250, not $25. The dashboard is two zeros off. If you ever see this tool open on a friend's monitor, hand them a calculator first.
English
3
6
15
2K
Rulya
Rulya@Rulyaxd·
Every agency still charges $10,000 for a "premium website." A guy just built a 3D animated brand site with a rotating jar and a checkout page. In an afternoon. From a chat window. The article shows the exact process. Six steps. Three hours. Zero code touched by hand. Step 1 is a prompt. Step 2 is a full HTML file. Step 3 is CSS from plain English. Steps 4-5 add JavaScript features and mobile fixes. Step 6 is Netlify: drag the folder, get a live URL in sixty seconds. The video shows what happens when the process is dialed. Tikka Masala brand site. 3D product spinning in the hero. Orange color grading. Section transitions. A "Let's Get Cooking" CTA button. A boutique digital agency's ask for something like this: $28,000-40,000 and a two-month timeline. The builder used one Claude Max subscription, one afternoon, one browser tab. Netlify free tier for hosting. Total spend to publish: zero. Old world: three-page proposal, a project manager, four calls with the client, a Figma deliverable, a dev handoff, two rounds of revisions, deploy day scheduled two weeks out. New world: type what you want, paste the code, refresh the browser. The kids replacing agencies right now aren't designers or devs. They're people who understand that describing something in plain English to Claude and then Netlify-dragging the result is a business. The agency that used to charge $10,000 for this is currently updating its own website. On WordPress. Wait until they see Fable 5.
⁠Durektor97@Durektor97

x.com/i/article/2073…

English
0
2
5
700
Rulya
Rulya@Rulyaxd·
@Zyron5m the gym membership perfected both at the same time
English
0
0
0
7
Zyron
Zyron@Zyron5m·
@Rulyaxd every subscription is betting you'll forget to cancel. Every box is betting you won't need to
English
1
0
1
18
Rulya
Rulya@Rulyaxd·
He canceled Claude Pro, ChatGPT, Cursor, Perplexity, and two writing tools in one afternoon. The reason: a $35 power bank. Same Mac mini he already owned. Same 32B local model he already had. What changed was that the power bank turned it into a four-hour untethered workstation, and once his AI stack could run on any coffee table, the monthly subscriptions stopped making sense. The stack he killed lands at $110-340 a month. $3,600 a year at the top end. His Mac mini cost $1,499 once. Power bank was thirty-five bucks. Payback five months. Every year after, $3,456 back in his account. Wireless keyboard on his lap. Vision Pro instead of a monitor. Sits at any table with any WiFi, or none. His model doesn't need the internet to answer him. Old world: rent GPUs by the token, hope the API doesn't rate limit you at 4pm, pay $20 a month whether you opened the app or not. New world: quantized 32B model on a box you own, $9 a month in electricity, four hours untethered, no cap. Local isn't a replacement for frontier models. That lane is real. But 80% of the daily grind transcribing recordings, drafting emails, sorting inboxes, single-agent tasks on a schedule that job now fits in a jacket pocket. He owns the compute. The subscriptions still own everyone else.
Zyron@Zyron5m

x.com/i/article/2073…

English
3
1
10
1.7K
0x0981
0x0981@LostInContetxt·
@Rulyaxd That power bank is the real MVP. Suddenly, the whole setup is portable and actually useful.
English
1
0
1
10
Adel Bucetta
Adel Bucetta@adelbucetta·
@Rulyaxd the real unlock here is realizing that hardware still matters, and we've been letting our software focus overshadow this crucial aspect of buildouts
English
1
0
1
13
Rulya
Rulya@Rulyaxd·
Every founder in my feed is wiring agents into everything. Almost none of them are watching who pays for the compute underneath. $725 billion. that's not revenue. that's just what four companies Amazon, Alphabet, Meta, Microsoft are guiding to spend building this thing in 2026 alone. up 77% from last year. A central bank looked at that number and wrote it down as a warning. Said the whole buildout crosses $1 trillion this year. Bigger than the Apollo program. Bigger than the interstate highway system. One line item. Here's the part that actually got me. MIT ran the study everyone's afraid to run. 300 real enterprise AI deployments. 95% show no measurable return on the P&L. METR went further. Asked experienced engineers to predict how much faster AI would make them. they said 24% faster. Measured result: 19% slower. They still walked away convinced they'd been faster. That gap how fast people feel versus how fast they actually are is the whole market in one sentence. None of this means the tech is fake. I use these tools daily, an agent reads my codebase and ships the patch while i sleep. That part's real. What's not real yet is the math. $1 trillion in. $25 billion out, at the very top of the stack. The tools work. the spreadsheet doesn't yet. Is it the same shape as 1999, or different this time?
testorosso@test0rosso

x.com/i/article/2072…

English
0
0
1
56
Rulya
Rulya@Rulyaxd·
The biggest AI agent guide making the rounds right now isn't about smarter prompts. It's about building something that keeps running after you close the tab. Pause at 0:14 . watch the coaster mid-drop for the "drop tested" shot. That's not a product photo that's a full render Claude scripted straight into the page. Three ways in: - agent services: scope one repetitive task for one client, build a working prototype against their real data, charge for the outcome. fast first dollar, caps at your hours - agent templates: same logic packaged once, sold on gumroad or your own site to buyers with the same problem. no client calls, no ceiling - automation agency: both combined into retainers. slowest start, compounds the longest The guide's own line for it: an agent that reconciles refunds against payouts every morning sells itself. An agent that "does everything" sells to no one. No coding required for most of it. connectors + scheduled tasks + tool use went from an engineering project to a weekend build. The actual bottleneck isn't the tech anymore. It's picking the one task worth automating first
yurshev@yurshevv

x.com/i/article/2070…

English
1
0
3
99
Rulya retweetledi
Wise
Wise@trikcode·
Fable 5 is not fable 5 anymore - they lifted export controls after adding new classifiers - added new safeguards. - and now the routine coding/debugging falls back to Opus 4.8
Anthropic@AnthropicAI

Claude Fable 5 will be available again globally tomorrow. After a series of productive conversations with the US government, we're redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding and debugging will fall back to Opus 4.8. We’ll continue to refine these classifiers over the coming weeks to reduce false positives and better distinguish genuine misuse from legitimate requests. We’ve also begun drafting a consensus framework—with Amazon, Microsoft, Google, and other Glasswing partners—for assessing the severity of AI jailbreaks and how AI developers should respond to them. We invite other industry partners and model providers to join us in this effort. Finally, we’re scaling up our collaboration with the US government on model testing and safeguards. This will include pre-release access to models and safeguards for evaluation, information sharing on jailbreaks and misuse, and dedicated resources for joint research. Thank you to our users for your patience, and to our partners across the government, industry, and the research community who worked alongside us to make Fable 5 available again. Read our full blog: anthropic.com/news/redeployi…

English
102
45
967
331.6K
Rulya
Rulya@Rulyaxd·
@Nezukoa4 most people would have sold at the bottom to stop the bleeding he went to investors instead and promised not to let them down
English
0
0
0
15
Nezuko
Nezuko@Nezukoa4·
Jon Gray bought Hilton for $26 billion at the very peak before the 2008 crisis, wrote the investment down by 71%, and by his own words should have been thrown out, yet ended up making $14 billion on it, the most profitable deal in the history of real estate private equity "by every measure, I shouldn't be sitting here. I should have been carried out" "I remember going to investors, one of our state pension funds, and the man almost got physically sick. I walked out of the meeting and thought: I can't let him down" "in early 2009 I told Chris: the good news is, it can't get any worse" in 18 minutes you'll learn how not to repeat his mistake with debt and what he did to survive ↓
Nezuko@Nezukoa4

Wall Street credit investor Scott Goodwin, managing $30 billion at Diameter Capital, broke down what is happening in private credit. "Private credit is a $40 trillion market. Today I will only talk about the $2 trillion slice - leveraged finance / direct lending" "Many portfolios have 30-40% in SaaS… and they are underreporting these numbers" In 18 minutes you will understand why private credit grew to $2 trillion, why the money went into SaaS, how AI is affecting these loans, and where the BDC risks are right now.

English
14
2
53
23.7K
Rulya retweetledi
Claude
Claude@claudeai·
Introducing Claude Science, a new app designed with every stage of research in mind. Artifacts traced to their code, environments managed on demand, and 60+ optional scientific databases that you can connect. Available now in beta.
English
1.1K
3.5K
35.9K
9.6M
Rulya retweetledi
Maxsim
Maxsim@OddsLedger·
He turned Claude AI into a 10 channel YouTube empire - and it's printing $24,937. A developer spent 6 months building a system that runs 10 faceless YouTube channels on autopilot - now generating $24,937/mo combined. No camera. No face. No manual editing. The entire pipeline is built around Claude AI: research, scripting, voiceover, even thumbnail generation - all automated. Here's the exact step by step system behind it, from blank page to posted video.
English
0
3
11
207
Rulya retweetledi
Rulya
Rulya@Rulyaxd·
A guy wired a $1,499 Teenage Engineering TP-7 into Claude Code. Now he ships landing pages by talking to a beige tape recorder on a kitchen table. The terminal says "listening." He talks. Opus 4.8 with a 1M context window starts drafting a landing-page redesign before he finishes the sentence. He taps the play button on a Walkman lookalike and the agent reads his codebase, lists the project directory, asks clarifying questions back without him touching the keyboard once. The stack: TP-7 as the mic. A CTRL USB module routing the audio. Claude Code in auto mode. A $200 Claude Max subscription doing the actual work. Senior frontend at a Series B agency goes for $185k a year, four interview rounds, a two-week take-home. He's doing the same work with his hands off the keys. Pause at 0:38. He types "Hi, welcome to my homepage." Claude appends the Mandarin translation in the same line. He didn't ask for it. The agent guessed his next ten minutes of work. The TP-7 was designed for field musicians who tape interviews on the move. He bought it to talk to a model. Vibe coded into a workflow most devs are still grinding LeetCode for. Git worktrees so three Claude sessions can run in parallel while one is still listening. The interface war is over. Nobody noticed the winning interface looked like a 1979 cassette deck.
English
0
1
0
141
Rulya
Rulya@Rulyaxd·
@eng_khairallah1 guessing what you wanted vs delivering exactly what you needed that gap is 25 settings nobody bothers to find
English
0
0
0
35
Khairallah AL-Awady
Khairallah AL-Awady@eng_khairallah1·
this is f*cking gold Andrej Karpathy joined Anthropic five weeks ago. A friend on his team just showed me 25 Claude features that most users have never touched. I configured all of them. The very first message was different. Not slightly different. A completely different tool. Claude stopped guessing what I wanted and started delivering exactly what I needed. Every single time. Bookmark this before it gets buried in your feed. Read it now, then check the article below.
Khairallah AL-Awady tweet media
Khairallah AL-Awady@eng_khairallah1

x.com/i/article/2068…

English
24
24
123
20.9K
Rulya
Rulya@Rulyaxd·
@milesdeutscher 60 seconds to connect, then it researches while you sleep
English
0
0
0
45
Miles Deutscher
Miles Deutscher@milesdeutscher·
Turn Claude Code into your personal finance assistant in <60 seconds. It can research trending AI subsectors (robotics), analyse tickers like $SPCX, conduct deep research while you sleep & more. All directly in your terminal. Here's how to connect it in just <60 seconds: 1. Open Claude Code and paste: "claude mcp add --transport http financial-datasets https:// mcp. financialdatasets. ai/" 2. Authenticate Type: "/mcp" inside Claude Code and complete the OAuth flow in your browser. You can verify the server is connected at any time: "claude mcp list" 2. Start prompting Example prompts: “What is SpaceX's current market cap?" "What's a good entry price on $SPCX?" “Show me Tesla’s revenue for the last 4 quarters.” 3. Help If you run into errors, just ask Claude to help you by scanning the official docs here: #claude-code" target="_blank" rel="nofollow noopener">docs.financialdatasets.ai/mcp-server#cla…
English
42
17
244
60.5K
Rulya
Rulya@Rulyaxd·
@antpalkin the agent grading its own work will always say it’s great that’s why the real upgrade is a second agent whose only job is to say no
English
0
0
0
5
cvxv666
cvxv666@antpalkin·
Anthropic team quietly stopped prompting their AI months ago. They just dropped an 11-page playbook on what they do instead - and one trader already turned it into $377,000. "Loop Design - The Playbook for Agentic Systems" The idea is simple: you don't type a request and wait. You build a loop that runs on its own - it finds its own work, checks it, and repeats. You stop being the one driving. The biggest point in the paper: an AI that grades its own work will almost always say it's great. So the most important piece isn't a smarter agent. It's a second one whose only job is to say no. That's exactly what this trader built for money. His Bitcoin bot treats every strategy it finds as fake until a separate check proves it's real. Last round it threw out 10 of its 12 "winners" and kept the 2 that held up. $377,000 so far. Read the paper, then see how he turned it into money below.
cvxv666 tweet media
cvxv666@antpalkin

Wall Street burns billions trying to predict Bitcoin. A 28-year-old self-taught coder in Warsaw made $377,000 by not even trying. He'd lost money on three trading bots before this one. Each looked perfect on paper, then started losing money the moment he ran it for real market. So he built bot that doesn't trust itself. His wallet: @easypredict?via=cvxv666" target="_blank" rel="nofollow noopener">polymarket.com/@easypredict?v… The truth is simple: you can't predict the next five minutes of Bitcoin. It's a coin flip. Anyone selling you a "prediction" is selling you nothing. So he stopped predicting. The bot hunts the moments the crowd is wrong instead. Here's the part that makes the money, and it's the opposite of what everyone builds. Any strategy can be made to look amazing on past data. On a 5-minute chart, most of them are just lucky, not real - and they stop working fast. So the bot treats every strategy it finds as fake until it proves otherwise. Each one has to pass a hard test: > test it on old data → test it on data it's never seen → try to break it on purpose → cut it down to the one thing that matters → run it forward → keep it only if it still works Last round, 10 of its 12 "winning" strategies turned out to be fake. It kept the 2 that actually worked and dropped the rest. And it never stops - building, testing, and dumping strategies around the clock. What worked yesterday can stop working today, so the second one starts losing, it gets cut before it costs you a thing. The result: $433,000 across 2,955 trades All his old bots tried to be right. This one just tries to catch itself being wrong - and that's why it's still alive. Bookmark this article below - it's the breakdown that explains why your last bot died. It's pruning and trading right now. Copy its wallet and skip to the edges that survived: @cvxv666" target="_blank" rel="nofollow noopener">kreo.app/@cvxv666

English
22
6
75
4.5K
Rulya
Rulya@Rulyaxd·
@gippp69 40 minutes a day for this kind of output is wild
English
1
0
1
818
Gipp 🦅
Gipp 🦅@gippp69·
THIS GUY TURNED HIMSELF INTO AN AI GIRL IN ONE SECOND. THE WORKFLOW BEHIND IT CAN RUN A $5,000/MONTH FANVUE PAGE IN 40 MINUTES A DAY 00:01 he raises his arms and instantly turns into an AI girl sitting in the same chair, ready to stream without ever showing his real face Claude creates the name, personality, captions and replies. one prompt can generate a full week of content while keeping the same character across TikTok and Fanvue ComfyUI with Flux builds the face and photo library. Kling 3.0 animates the images, then CapCut turns them into 30 to 50 short videos in one afternoon three daily streams and one clip reaching 400,000 views can send 70 buyers into a $15 Fanvue subscription. paid messages and tips are what push the page past $5,000 he stays behind the camera, Claude runs the brain and the AI girl becomes a business that works every day
Insomnia@insomnia_vip

x.com/i/article/2071…

English
53
69
632
126.4K
Rulya
Rulya@Rulyaxd·
@codewithimanshu 78% win rate over 3,210 trades is the part that actually matters
English
0
0
0
111
Himanshu Kumar
Himanshu Kumar@codewithimanshu·
This guy made $1,003,045 trading crypto with an AI bot. Not from one lucky trade or a bull market. From 3,210 trades executed over 1 years using the same data-driven system. 78% win rate. His biggest winners turned small positions into massive payouts: $34 became $4,113 $252 became $22,202 $46 became $4,048 $138 became $7,148 $63 became $3,259 The interesting part? The bot does not try to predict where BTC will go next. It looks for moments when the market is pricing something incorrectly and acts before the crowd catches up. No emotions. No FOMO. No panic selling. Just probabilities, risk management, and execution. Most traders spend years trying to remove emotions from their decisions. An AI bot starts with no emotions in the first place. A million dollars was not made from one trade. It was made from thousands of small edges compounded over time. The same framework: → Scan for mispricing across markets → Size positions using Kelly criterion → Execute in milliseconds before repricing → Let probability play out over volume You can test this style of trading yourself with 48 hours of free paper trading. See the signals in action without risking a dollar. You only need Claude + device + 1 hour to deploy. Giving This Free for 24 hours. To get it: 1. Comment the word Crypto 2. Like and Retweet this post 3. Follow me @codewithimanshu (so i can DM you) Save this post. Deploy the bot this weekend. Start with paper trading. Scale on evidence.
English
192
196
296
40.1K
Rulya
Rulya@Rulyaxd·
@RetroValix keeping up+down combined under $1 and just repeating that edge thousands of times is the most boring sounding strategy that actually works, which is usually how you know it’s real
English
0
0
0
32
VALIX
VALIX@RetroValix·
This guy built a trading bot with Claude and made $96,000 on Polymarket with an average trade size of $3 His bot trades crypto Up/Down markets using a hybrid strategy that combines two-sided market making, arbitrage, and directional trading His strategy has five steps: 1. Monitors multiple crypto markets 2. Calculates its own fair probability for Up and Down 3. Places orders at prices that keep the combined cost of Up + Down below $1 4. Manages the unhedged portion of the position 5. Keeps directional exposure when its model identifies one outcome as undervalued This guy’s Polymarket account: @neversmiling?via=670" target="_blank" rel="nofollow noopener">polymarket.com/@neversmiling?… Using this strategy, he captures a small edge, repeats it thousands of times, and steadily grows his deposit
VALIX@RetroValix

x.com/i/article/2070…

English
32
75
324
84.3K

Keşfet

@Zyron5m @LostInContetxt @adelbucetta @Nezukoa4 @eng_khairallah1 @milesdeutscher @elonmusk @BarackObama