100F.exe

907 posts

100F.exe banner
100F.exe

100F.exe

@100F_exe

Stop doing manual work. build systems. let AI scale everything.

参加日 Ocak 2022
342 フォロー中403 フォロワー
100F.exe
100F.exe@100F_exe·
Free prompt library for Claude from Anthropic that'll make working with AI way faster and save you tokens 🔵 Jump into the library and pick the category you need 🔵 Grab a ready-made template and tweak it for your use case 🔵 The prompts are optimized for AI, so you burn fewer tokens 🔵 Spend less time writing prompts and learn how to craft better ones yourself Bonus: here's a hack for anyone trying to cut Claude costs by up to 60%. The new Fable model is insanely good at reading text from images and uses around 60% fewer tokens compared to feeding it long text prompts. Bookmark the library and send this to your favorite vibe coder
100F.exe tweet media100F.exe tweet media
English
3
0
8
81
100F.exe
100F.exe@100F_exe·
@Frogiik Morocco keeps making history. African football is on another level right now
English
1
0
1
13
Frogik
Frogik@Frogiik·
🌍🔥 Morocco is the first quarter-finalist of the 2026 World Cup! Despite a difficult first half, the African team rallied and defeated Canada 3-0 💪 Canada is the first host country to leave the tournament
Frogik tweet media
English
1
0
2
58
Bober_smart
Bober_smart@Bober_smart·
A 20-year-old guy from India decided to prank his girlfriend, but the prank turned into a business with an income of $5,068 in two weeks With the help of Claude, he created an AI girlfriend and generated a video where they are kissing, which he supposedly sent to his girlfriend by accident The girlfriend couldn't calm down for a long time and didn't believe it was AI: it looked very realistic They decided to create a TikTok account for her, and after a week, 85,432 people had subscribed to the account They started thinking about how to make money from this and set up an account for her on Fanvue > TikTok for promotion > Fanvue for income After two weeks, 512 people signed up for a paid subscription -> subscription price $9.99 -> income $5,068 It started as a prank, but now it is a family business
Bober_smart@Bober_smart

x.com/i/article/2070…

English
13
1
27
558
Bober_smart
Bober_smart@Bober_smart·
@100F_exe That’s a real alpha move, buddy thanks for the smart content
English
1
0
0
22
Logics
Logics@immortalhowwl·
-$5,576,579.00 on FIFA WC 2026 after 12 polymarket predicts worst run in the project’s history today he’s betting $3,865,536 on Morocco WIN - NO Canada win - YES $717k Canada/Morocco draw $388k life teaches him nothing tag a gooner with his luck who could take his spot
Logics tweet mediaLogics tweet media
Logics@immortalhowwl

Salam men l-Maghrib who are u rooting for today?

English
13
0
28
869
100F.exe
100F.exe@100F_exe·
@Damir_Akaza Every legendary trader has a story that would violate their own risk rules today.
English
0
0
0
15
Damir Akaza
Damir Akaza@Damir_Akaza·
Steve Cohen at 21 traded without a hedge what he'd forbid anyone at his firm to do today, and he explains how it turned him into a legend. the founder of Point72 and owner of the New York Mets explains how he reads the markets and applies the same principles to running the Mets. "they hired me to do options arbitrage. but I thought: why hedge if you're getting the price right? I could sell it higher the next day" "most of the people I competed against in the 90s aren't in business today. they didn't adapt" "if I hadn't been flexible and adaptable, I wouldn't be here today" watch to the end to understand what separates the traders who stay in the game for decades from the ones who disappear ↓
Damir Akaza@Damir_Akaza

Marc Rowan grew Apollo 14 times over, outpaced Apple and Microsoft, and manages $650 billion. and when his partners expected a breakdown of some genius strategy, he got up and said: we just got lucky. The CEO of Apollo explains how money is actually made and why almost everything investors believe falls apart over the long run. "How do you know who was a good investor over the last 10 years? take out the last two. track record is useless" From 2008 to 2022 they printed $8 trillion, and everything went up, the good and the bad. three years ago these people were called geniuses, today they say they just got lucky, and half of them are out of a job. "In private equity you make 10 decisions a year at most, everything else is ceremony. buy a good company cheap and sell it higher" "80% of trading is the S&P 500. 10 companies are 35% of the index. the whole market is basically levered to Nvidia and Apple. active management has failed to beat the market 90% of the time for 20 years" and he predicts that in the next 10 years we'll see the death of active management. bookmark it and watch today ↓

English
8
1
44
6.1K
100F.exe
100F.exe@100F_exe·
@savipww I trust an agent with reminders, not my bank account yet.
English
1
0
1
14
savip.
savip.@savipww·
everyone's out here hiring co founders I just built mine in under 3 minutes here's the actual build open Claude Code or Codex, I use Claude Code create a new local folder, this is where the agent lives go to 🔗hellotrillion,ai/p/start-here, they give you a ready made prompt paste that prompt into Claude Code, switch to plan mode, hit enter it asks you a series of questions to shape the agent around your actual goals then go to ElevenLabs for the voice, plans start at $5 grab the voice ID, Claude Code needs it then Deepgram for speech to text, grab that API key too paste both back into Claude Code, answer whatever it asks next run it on localhost and you're talking to your own AI co founder mine already told me its primary objective is making me a million in recurring revenue it also said it gives me six months, maybe five if I stay off TikTok an AI co founder that roasts you for scrolling is either the funniest or the most useful thing I've built this year would you actually trust an agent to manage your business chaos or is that a step too far? full step by step build documented, saving it below ↓ x.com/savipww/status…
savip.@savipww

x.com/i/article/2072…

English
5
1
14
231
100F.exe
100F.exe@100F_exe·
@kocer_eth The best workflows start before the first prompt
English
0
0
0
3
kocer
kocer@kocer_eth·
THIS GUY MADE CLAUDE COWORK REPLACE 60% OF HIS PROJECT WORK WITH A 30-MINUTE SETUP Not by finding a magic prompt. By giving Claude memory before giving it work. Ray Fu’s demo is basically this: 1. Create a Claude Context folder Inside it: about-me.md brand-voice.md working-style.md That folder tells Claude who he is, what he is building, how he writes, and how he prefers decisions to be made. So every session does not start from zero. 2. Put the same operating context into Claude settings This is the part most people skip. They keep prompting Claude like it is a stranger. Ray makes Claude load the working rules before the task even begins. 3. Add role-specific tools The video shows sales, marketing, legal, and data-style plugins. The point is not “more tools.” The point is giving Claude a seat: researcher marketer data analyst legal helper sales assistant A general chatbot becomes more useful when the job is defined before the request. 4. Make Claude ask questions first The line is simple: “Ask me clarifying questions before you start.” That changes the output immediately. Instead of writing a generic landing page, Claude asks about the product, audience, desired action, brand guidelines, and offer first. That is the payoff. The 60% replacement claim is not really about Claude doing someone’s whole job. It is about deleting the repeat work: re-explaining context reloading research setup fixing tone again restating preferences watching the model guess before it understands The reusable system is boring: context files global instructions role tools back-prompting Most people keep asking AI to work faster. This setup makes AI understand the work sooner.
Francesco@francescoinweb3

x.com/i/article/2073…

English
5
1
16
526
Rezzi
Rezzi@Rezzi_sol·
Someone turned a $150 ex-crypto miner into a local AI machine worth watching. The board is the AMD BC-250, a weird piece of surplus PlayStation 5-class hardware that went from crypto mining board to e-waste, then cheap Linux gaming machine, and now local inference experiment. This is not interesting because it beats a clean RTX 4090 setup. It does not. It is interesting because the local AI community keeps finding ways to squeeze useful performance out of ugly surplus hardware when the price gets low enough. The BC-250 gives you 6 Zen 2 cores, 24 RDNA 2 CUs, 16GB shared GDDR6 and around 448GB/s memory bandwidth. With the annoying parts included, the total build cost can land around $150-250. That is where the payoff starts. Small local models stop feeling like a Raspberry Pi science project and start feeling like something you can actually tinker with on a desk machine. The performance curve is the best part. The TechMakesArt video shows Ollama starting around 25.67 tok/s, then llama.cpp pushing it to 37 tok/s, more tuning reaching 54.99 tok/s, and batch serving hitting 174.95 tok/s on one node. Same strange board. Different software path. Completely different result. The tradeoff is that this is not a normal consumer PC. It is Linux-only, Windows GPU support is not clean, power can sit above 220W, and you may end up touching BIOS settings, cooling, GPU governors, Vulkan builds, forks and community docs before it behaves. So BC-250 is probably not the best local AI box for most people. It is a reminder that the next cheap local AI PC may not look like an AI laptop. It may look like dead miner hardware that a community refuses to let die. For plug-and-play, buy something boring. For cheap local inference with people already squeezing tokens out of it, BC-250 is suddenly worth watching. Dead miners, live compute.
Rezzi@Rezzi_sol

This guy built a home AI supercomputer out of Mac Studios. Five machines, wired together, running as one local inference cluster through Exo. The point is not that it looks insane on a desk. The point is what it replaces. Most people rent intelligence one message at a time. ChatGPT Pro. Claude Code Max. Cursor. Transcription tools. Research tools. API credits. The stack quietly turns into a $400/month bill just to use compute sitting in someone else’s datacenter. This setup flips the model. Macs use unified memory, so the GPU can use the same memory pool as the system. Exo then splits large open-source models across multiple machines, letting the cluster run models a single computer could not handle alone. DeepSeek. GLM. Kimi. Qwen. The kind of models people usually access through cloud APIs can now sit on a desk, running privately, overnight, with no rate limits and no token anxiety. The old flex was paying for every frontier model. The new flex is owning the box that keeps thinking after you close the tab. Bookmark this and read the full setup below.

English
5
0
15
253
bnd
bnd@No13121Bnd·
For me, and perhaps for many others, it feels like the approach implemented by @pmfi_cc offers a new perspective on interacting with the Vault Sector. However, that does not mean it should not be studied or analyzed in depth. Much like the concept of "yield," this project can also be seen as the beginning or even the continuation of a new era for Founders. With their skills, they are now able to develop projects almost single handedly, delivering not only finished products but also proving their willingness to work around the clock. This demonstrates to both themselves and crypto users that the class of decentralized assets is becoming more and more accessible across a broader range of opportunities. How it works: 1. Deposit USDC - Deposit USDC on Base. A vault report processes the request and mints pARB at the official price per share. 2. Hold pARB shares - Your pARB shares track the vault's price per share as captured spreads settle. 3. Automated execution - The vault buys YES and NO whenever their combined cost is below $1 after fees 4. Redeem to USDC - Request redemption and pARB is priced to USDC at the official PPS. Requests are processed by report in FIFO order, subject to vault liquidity. The most important question is: where does the edge come from? A single outcome trades on @Polymarket and @Kalshi at the same time. Buy YES on one venue and NO on the other for a combined 97¢, and at settlement exactly one pays out $1 - the 3¢ gap is the captured edge, independent of who wins. CA: 0xD77CE6D3137342Bb5174673bDab5f51Db16fcBA3
PMFI@pmfi_cc

x.com/i/article/2072…

English
2
1
15
216
Kira
Kira@kiira7094·
My predictions for the 2026 World Cup at @Polymarket today: I expect both favorites to advance, but neither match should be straightforward. 1. 🇨🇦 Canada vs. 🇲🇦Morocco Both teams to score Morocco should go through, but Canada won’t sit back and wait to lose. Their aggressive style can expose them defensively, but it should also create enough chances to score at least once. 2. 🇫🇷 France vs. 🇵🇾 Paraguay Under 3.5 goals France are the clear favorites. Paraguay, however, are extremely defensive and will likely spend most of the match trying to slow them down. France should win. Just don’t expect a goal fest. NFA. DYOR.
Kira tweet media
Kira@kiira7094

My today's @Polymarket combo Yesterday’s combo hit. Today, I’m betting against two favorites winning comfortably. 1. 🇪🇸 Spain vs. 🇦🇹 Austria Austria +2.5 Spain are the better team, but are they three goals better? Austria are ranked inside the top 20 and have performed well at this World Cup. Spain, meanwhile, haven’t lived up to expectations. 2. 🇵🇹 Portugal vs. 🇭🇷 Croatia Croatia +1.5 Portugal haven’t looked convincing either. Their only win came against a weak Uzbekistan side, and now they face Croatia’s strong defense. Portugal might win, but a two-goal victory feels unlikely. I don’t need any upsets today. I just need both favorites not to run away with it. NFA. DYOR.

English
18
4
47
2.8K
Murmur
Murmur@armouredme·
Iran won't fully close it's airspace Low risk trade on Polymarket polymarket.com/event/iran-ful… July 15 priced at 4% July 31 at 10% $576K and $245K traded Here is what those numbers are actually pricing The airspace is currently mid-reopening under the US-Iran deal Eastern side already back open Western sector still restricted Late June saw a fresh strike exchange break the calm for a few days Both sides pulled back June 29 Nothing since Full shutdown has only happened twice this year Both times during live strikes, not just high tension Right now sits in the second category, not the first Khamenei's funeral runs July 4 through 9 Defense systems stay on alert the whole stretch That is the actual risk window here A funeral is not the same trigger as an active strike cycle 4% on July 15 reflects that gap correctly 10% on July 31 leaves room for something breaking before then The bar is a total FIR shutdown A partial NOTAM like the one in late May does not clear it
Murmur tweet mediaMurmur tweet media
English
11
1
45
635
Bober_smart
Bober_smart@Bober_smart·
A 24-year-old guy from Japan was dumped by his girlfriend and decided to take revenge on her, but the result will surprise you: he earned $3,246 in 2 weeks He created an AI girl named Claude, generated photos and videos He just wanted to generate photos with a girl and post them on Instagram so that his ex would regret it The girl turned out to be so realistic that he decided to start a TikTok for her, and in just a week the account gained 65,328 new followers Men started writing to her in direct messages, and he realized that he could make money on this: he created an account for her on Fanvue > TikTok is traffic > Fanvue for profit In 2 weeks, 325 people subscribed to the paid subscription -> paid subscription $9.99 -> profit $3,246 Often what seems not serious turns into something more
Bober_smart@Bober_smart

x.com/i/article/2070…

English
30
3
70
8.3K
Riley West
Riley West@rileywestreel·
As a student, Carl Icahn lost a full week's pay in a poker game to the owner of the beach club where he worked. Two weeks later, after reading three books on poker strategy, he was winning $500 a week and ended the summer with $2,000, against the $750 he needed for room and board at Princeton. "The real or liquidating value of many American companies has increased markedly in the last few years; however, this has not at all been reflected in the market value of their common stocks. Sizable profits can be earned by taking large positions in undervalued stocks." From 1968 to 2011, he compounded an initial $100,000 at a 31% annual rate. Over the same period, Buffett's Berkshire Hathaway grew at 20% a year. "The consensus thinking is generally wrong. If you go with a trend, the momentum always falls apart on you. So I buy companies that are not glamorous and usually out of favor." In October 2012, Netflix shares had crashed 80% to $58, and Icahn put in $321 million for nearly 10% of the company. By his own account, the position was up 457% in 14 months, and over three years Netflix earned him close to $2 billion. See below ↓
Riley West@rileywestreel

Carl Icahn. He's 90, and he still personally owns 86% of Icahn Enterprises. "You have to buy things where the rest of the world are looking at you and think you're a little bit crazy" February 2026: entities linked to Icahn bought 783,404 more shares of CVR Energy for $16.4M, pushing his CVI stake past 71.2M shares. CEO Andrew Tino on the Q1 2026 earnings call called him "a living legend of activism" $IEP trades around $7.40 while its dividend yield sits at 27% annually.

English
5
1
13
658
Zentrix⌚️
Zentrix⌚️@ZentrixHQ·
📖WHY CHANGING THE LENS MATTERS MORE THAN CHANGING THE SHOT Five lens decisions made in advance beat five prompts written in isolation. Most AI video sequences use the same virtual focal length across every cut, and the result feels flat even when the individual shots look good. Nothing changes in scale or intimacy from one moment to the next, so the sequence reads as a series of clips rather than a scene with rhythm. A shot list that varies the lens on purpose fixes this before a single prompt gets written. ▪ Rack-focus close-up on a longer lens (85mm) — compresses the background and isolates the subject, used for the emotional beat that needs full attention ▪ Handheld medium on a standard lens (35mm) — keeps some environment visible while staying close, used for the moment that needs presence without full isolation ▪ Wide orbit on a shorter zoom (24-70mm) — reveals scale and context, used to place the emotional beat inside the larger scene ▪ The order matters as much as the lenses — tight to loose (or loose to tight) creates a felt sense of reveal or focus that a flat sequence of same-length shots never produces ▪ Each lens choice gets locked into the shot list before generation, not decided per prompt, so the rhythm is intentional instead of accidental This works the same way regardless of the scene. A proposal, a chase, a product reveal — the lens pattern is what turns disconnected clips into something that feels directed. The subject changes. The structure doesn't. 📥 tomorrow's post covers how to keep lighting and color palette consistent across a five-cut sequence shot on different lenses
Zentrix⌚️@ZentrixHQ

x.com/i/article/2069…

English
3
1
17
692
bnd
bnd@No13121Bnd·
I am experimenting with a charming narrative, because the autonomy of AI agents for me is not only about code, but also about discovering beauty in the details of prompts. Shinkai is a free open-source app for creating local AI agents without coding. @ShinkaiLocalAI supports: > Local models through @ollama > Cloud models > MCP > Crypto payments (configuration options for x402)
bnd tweet media
Shinkai: Local AI Agents@ShinkaiLocalAI

If you've been putting off setting up a local agent because it seemed like too much: This weekend is the time. Download: 2 minutes. Setup: 10 minutes. First working agent: 20 minutes. The hard part is deciding what to automate first. Not the setup. 🐙

English
4
0
15
419