Gofra
228 posts
Gofra
@Gofralo
Just a simple guy, who loves probabilities
Katılım Nisan 2012
55 Takip Edilen57 Takipçiler
The wildest part isn't the $3,200/month.
It's that this kid never had to be on camera, never had to build an audience first, and never had to know more than the people watching — just slightly ahead of them.
That's the actual unlock AI created: you don't need to be the expert anymore. You need to be fifteen minutes ahead of someone who hasn't opened the tool yet.
Most adults are sitting on more "fifteen minutes ahead" knowledge than they think. What's yours?
Gofra@Gofralo
A 17-year-old runs a faceless YouTube channel about AI tools. No camera, no studio, no face on screen. $3,200 a month from ad revenue and affiliate links, alongside school. The channel doesn't review anything fancy — it just explains AI tools other people are too lazy to figure out themselves, then links to them. The lesson isn't "be 17." It's that the entry barrier moved from "be charismatic on camera" to "be willing to explain one tool clearly." What's one AI tool you actually understand well enough to explain to someone else in two minutes?
English
A 17-year-old runs a faceless YouTube channel about AI tools. No camera, no studio, no face on screen.
$3,200 a month from ad revenue and affiliate links, alongside school.
The channel doesn't review anything fancy — it just explains AI tools other people are too lazy to figure out themselves, then links to them.
The lesson isn't "be 17." It's that the entry barrier moved from "be charismatic on camera" to "be willing to explain one tool clearly."
What's one AI tool you actually understand well enough to explain to someone else in two minutes?
English
Most AI agents start losing the thread by tool-call #8 or #9 in a long task.
This setup stays coherent through 30+ calls in a row — read the file, follow every link, check the tags, scan for duplicates, write the report — without forgetting what it was asked at call #1.
The trick isn't a smarter model. It's letting it see your entire notes graph at once instead of feeding it fragments one chat at a time.
Anyone else still resetting their AI session every 20 minutes because it loses the plot?
Mr. Buzzoni@polydao
English
Xiaomi has released its own free AI tool for vibe-coding ➜ MiMo Code
The most exciting part? It comes bundled with a free model called MiMo Auto. No API keys required to get started. You can easily connect APIs from virtually any popular LLM provider and use your own models
Key features include:
> Persistent memory across sessions
> Sub-agents that automatically spawn and work in parallel
> /dream — saves useful knowledge from the session into the project’s long-term memory
> /distill — turns repeated actions into reusable skills
> /goal — uses a dedicated verifier model to determine whether the task is truly complete
Installation:
Mac / Linux:
curl -fsSL mimo.xiaomi.com/install | bash
Windows:
npm install -g @mimo-ai/cli
Launch:
mimo
100% open source
beamnxw ./@beamnxw
English
@Mahaximus_ the infrastructure race is where the real long-term money goes
English
They are putting DATA CENTERS in the ocean now.
Panthalassa, a startup from Portland, just raised $140 MILLION
what they do: build floating platforms that sit out at sea and run AI. no power grid needed. the ocean waves make all the electricity, the seawater keeps the chips cool
How it works: big floating balls bob up and down with the waves. that motion makes power. the power runs the AI chips inside
backed by PETER THIEL. Company now worth almost $1 BILLION
land is running out of room and power for AI, so the next move is simple - go to the sea.
English
@thegreatest_sv same prompts different results, this is the benchmark that actually matters
English
CHINA JUST DROPPED GLM 5.2 - A CLAUDE OPUS KILLER.
I tested three new AI models on the exact same coding challenges:
• GLM 5.2
• Kimi K2.7
• Claude Opus 4.8
Same prompts.
Same tasks.
Same rules.
The result shocked me.
GLM 5.2 won 4 out of 5 tests.
It built:
> A better endless runner
> A better product landing page
> A better simulation
> Better interactive experiences overall
And that’s not even the craziest part.
It comes with:
> 1M token context
> Agent support
> Lower pricing
The models coming out of China right now are improving at an absurd pace.
I don’t think most people realize how competitive this race has become.
Z.ai@Zai_org
Introducing GLM-5.2: Frontier Intelligence, Open Weights - Significant improvements in coding and agentic tasks - Strong long-horizon capabilities with a 1M context window - Two levels of reasoning effort: GLM-5.2 (max) pushes the limits, while GLM-5.2 (high) strikes a strong balance between performance and token efficiency - MIT-licensed open weights - Same API pricing as GLM-5.1 Tech Blog: z.ai/blog/glm-5.2 Weights: huggingface.co/zai-org/GLM-5.2 API: docs.z.ai/guides/llm/glm… Coding Plan: z.ai/subscribe Chat: chat.z.ai
English
@adiix_official entire games from prompts and we're still in early innings
English
You could literally:
- open GLM 5.2
- prompt 1: “build minecraft from scratch”
- prompt 2: “add caves, hostile mobs, crafting table”
- wrap the output in Electron
- list on itch.io for $4
- stack 3 of these a weekend
one kid + one evening = three shippable survival MVPs
Why are people still not doing this?
AdiiX@adiix_official
GLM 5.2 just killed Claude Fable 5. Watch 6:36 first if you only have a minute. That’s where it stops being a tech demo and starts being a game. One small sentence. Two prompts. The AI built Minecraft from scratch. Not a clone. Not a demo. A working game: - inventory + crafting table - ore mining: coal, iron, gold - underground caves - mobs: zombies, sheep, chickens, cows, pigs - day/night cycle - fall damage - locked 60 FPS The numbers are insane: Notch built the original Minecraft solo over months. Microsoft bought it for $2.5B in 2014. GLM 5.2 shipped a working copy in the time it took the creator to grab a coffee. The science is wild too: The model iteratively builds the systems itself. Inventory first. Then crafting. Then mining. Then mobs. No extra instructions in between. The grift opportunity is even wilder: Indie studios charge $50k+ for a survival game prototype. Unity freelancers sit at $80/hr. Now one kid + one prompt + one evening = a shippable MVP. Full timestamps: 0:25 - Minecraft in ONE prompt 6:36 - Minecraft in TWO prompts (caves, mobs, crafting) 9:40 - the verdict: GLM 5.2 or Claude Fable 5? Claude Fable 5 still holds up. But GLM added caves and hostile mobs nobody asked for. Save this before your knowledge cutoff stops mattering.
English
@hanakoxbt model deciding what context loads is a bigger deal than it sounds
English
Anthropic Product Lead:
"Most people pay full price for context they don't need. We rewrote the loop so the model decides what loads and what stays."
In 25 minutes Brad Abrams takes one CEO dashboard from $10 per run to almost nothing, live on stage.
The agent defers 100+ tool declarations until it actually needs one. Claude writes Python on the fly to pull only the bytes it needs from giant results.
Sonnet handles execution, Opus sits as advisor and catches the line Sonnet missed in the Gong transcript.
Watch the full talk, then save the setup below.
Hanako@hanakoxbt
English
Boris Cherny, Head of Claude Code:
"I don't prompt Claude anymore. I have loops running that prompt Claude and figure out what to do. My job is to write loops"
The best AI builders on Earth already quit writing prompts. They write loops - systems that run, check their own work, and go again until something survives
Peter Steinberger, who built OpenClaw, said same thing the same week: stop prompting your agents, start designing the loops that prompt them.
A prompt is one shot. A loop runs, checks its own work, and goes again until something survives.
I broke down exactly how that loop works - and who just handed it to retail.
cvxv666@antpalkin
ChatGPT can write you a thousand trading strategies from one sentence. 999 will quietly bleed your account dry. Every single one of them looks like a winner. One method pulls the real one out. Quant desks have run it for years. It's called loop engineering. It's not a smarter prompt. It's a loop: generate a strategy, test it, score why it failed, feed that back, and run it again until one survives data it has never seen. Man Group runs $150 billion. They built an AI that writes and tests hundreds of strategies a week. And second agent whose only job is to hunt the ones that secretly overfit and kill them before a cent goes live. > describe an idea -> backtest 5 years in 12 seconds > read why it died -> feed that back in > run it again until one survives data it never saw AI can write a thousand strategies. Loop engineering is the only thing that tells you which one was ever real. 89% of retail lost money in 2025. Almost none of it from bad ideas. All of it from trading the first guess and never running it through backtest loop. Horizon just put the whole thing in a chat box, free while the beta's open: join.horizon.trade/explore You're not losing to smarter traders. You're losing to the ones who opened the tab. Anyone can generate a strategy now. Almost no one runs the loop that proves it. That gap is the whole edge. You'll mean to check it later. Later is a "waitlist closed" page.
English
UPD: THIS GUY COULD HAVE WON $37 MILLION
If he knew about Polymarket combos…
I just realized the guy who won $7M yesterday left $30M on the table.
He traded the same take on 2 markets:
Meaning they either both win or both lose.
If he opened it as a combo, the odds would have multiplied.
As a result, he would’ve won WAY more:
($4.95M + $1.05M) * 7.14 (combined multiplier) = $42.84M
$42.84M - $6M (initial trade) = $36.84M
$37M vs $7M, guys.
This guy left $30M on the table because he didn't know about new feature.
I’m speechless.
Dexter's Lab@DextersSolab
🚨THIS GUY BECAME A MILLIONAIRE IN ONE DAY (he made just 2 trades btw) Fresh Polymarket account and $7.6 million profit. Address: [@grimdrip?via=dexter-molu" target="_blank" rel="nofollow noopener">polymarket.com/@grimdrip?via=…
] He predicted Czechia draw with 100% accuracy. And split the same take on 2 trades. Worth watching... English
Mustafa Suleyman:
"your workflows become your moat, not the AI model you use"
everyone has the same models now, what nobody can copy is how you trained an agent around your own process
Microsoft is building this with MAI models on their own Maia 200 chip and custom RL environments that tune agents around your workflows instead of renting intelligence
generic models are a commodity, owning AI that knows your business is the only edge left
full breakdown below 👇
leopardracer@leopardracer
English
@ventry089 design plus deployment is 2.5x leverage most designers still ignore
English
$7,000/month → $18,000/month. one year. one variable
UI/UX designer on upwork. january 2025 - 5 clients, design only. april 2026 - 12 clients, design + full code + deployment. 157% revenue growth
she didn't learn to code. claude code writes the code. she sells the working product. client gets the result. who built it - nobody asks
one of her clients spent 4 hours a day on manual data entry. one automation setup - 4 hours became zero. price tag - $1,500
upwork 2026: AI integration grew 178% year over year. chatbot development up 71%. freelancers using AI earn 47% more per hour
the freelancer charging $25/hour and the one charging $200/hour use the same tool. the difference - one sells time. the other sells a system
generic AI writing gigs are dying. AI micro-services that build working products - explosive growth. $500-$2,000 per chatbot setup
freelancing split in two. those who build with AI and those who compete against it
Pikachin@pikach_in
English
30% apr funding arb. nets 12% after costs. one slow exit and it's red
where the other 18% goes:
long spot, short perp, collect funding every 8h. 0.015% a period reads like ~19% a year
then the costs. taker fees round trip, 4-10 bps a leg, both legs, in and out. and funding doesn't hold. the rate slides toward zero in two days, so you bank $18 against $10-15 in fees and the "risk-free" trade sits flat or under
what nobody sizes for: only enter once a delta's held 5-7 days. one print isn't an edge
cross-exchange is the same trap. hyperliquid vs binance run 6-11% apr, prints have touched 48%. arbitrage should close that gap and it just doesn't. the capital that would close it is off chasing price
the yield is real. just thinner and slower than it looks
English
@0xTrackmind self-improving loops are the part nobody's ready to talk about
English
Creator of Claude Code:
"Almost 100% of our engineers are running 100+ agents with self-improving loops. Each run makes the agent better than the last."
This is the internal setup at the company that built Claude.
Not a tutorial. Not a course. The real thing.
Claude + loops + routines + dynamic workflows
1 hour. Free. Straight from Boris himself.
Bookmark this.
Horizon@horizon_trade_x
English
Jesse Livermore traded alone and went bankrupt four times
Every time because there was no one to challenge his thinking
Today the best traders aren't alone
They're in small private groups, where one says "I see a reversal" and another replies "no, here's the data."
Horizon just opened exactly that kind of place.
1000 traders. Free for now:
Horizon@horizon_trade_x
1,000+ traders already in the Discord Where strategies get built, backtested, and stress-tested -> discord.gg/horizontrade
English
Pixar's CEO after realizing a kid replaced his $200 million model with $124 and six tools
Fokki@0x_fokki
🚨 the clip you're watching was made for $1.20 Pixar burns $200 million and 4 years per film a 19-year-old spent $124 last month and made $12,345 > Claude wrote the script: 20 min > Midjourney built the frames: 40 min > Runway animated the cuts: 1 hr > ElevenLabs cast the voices: 25 min > Suno scored it: 15 min > Make shipped it to YouTube: auto every frame above paid for itself. the full pipeline is in the article above👇
English
One whale stacked USA Yes, Draw No, and Australia No simultaneously. Six figures. Same view three ways. While everyone copies him - I'm going the other way.
> Australia shut out Turkey. USA's attacking depth is shakier than the Paraguay scoreline suggests. The market drifted hard to USA and stopped thinking.
Australia plays structure and discipline - exactly the style that keeps games tight and makes draws happen at 22 cents.
The people copying this whale are going to watch their position bleed if Australia holds the line for 70 minutes.
// Australia +1.5 looks like the most ignored position on the board right now. Draw is live. The crowd is too busy following six figures to check the actual data.
I'm watching this through @Prophetzone - power ratings, path data, where sharp money moves before kickoff. Everyone else is watching the whale.
My position tonight: Australia +1.5 and Draw YES. Fading the crowd.
Atenov int.@Atenov_D
This free model just beat every closed-source AI on coding benchmarks. Open weights. 6x cheaper than Opus. The labs don't want you to know it exists. > GLM-5.2 from Zai just topped Code Arena - the first open-weights model to ever hold the #1 coding spot. Not a leaked weight, not a fine-tune. A fully open model beating GPT and Claude on their own turf. It's live on Hugging Face Inference API right now. 5 providers: Novita, Together AI, Fireworks, Deepinfra, Zai. OpenAI-compatible client. → Go to huggingface(.)co → grab HF_TOKEN from account settings → plug into any OpenAI-compatible client 6x cheaper than Opus. Companies bleeding on AI bills are already routing to this for orchestration, caching, and token optimization. The smart money moved before the headline dropped. > Now you know. huggingface(.)co/zai-org/GLM-5.2 Bookmark this before everyone else figures it out.
English