Phosphen
@phosphenq • 15,935 subscribers
Сhoose life $BODKIN CA: 0xb06b1e58f5ba2a3df1ab74c01cb2a44c5395b3be
Shorts
Videos
i fired my whole team and hired spiders seven Grok Bots on one computer, $20 a night, $675 of work shipped while i slept that is them on your screen, the tape from last night each bot owns one lane: - research briefs before the morning call - inbox zero by sunrise - draft packs for the content queue - PR reviews before anything merges every job they catch gets wrapped in silk and priced one of them wanted $184 for an ad test and the gate froze the money until my tap seven workers, zero payroll any task you pay a person to repeat, an agent can run cheaper, once you wire it right the full setup, plugins and locks included, below
Phosphen301,062 Aufrufe • vor 19 Tagen
She's 22. Her AI runs inside the Fortune 100. a16z just led her $21M raise. "Put up the most minimal fake version you probably can of something and then go sell it." In 17 minutes, MIT grad Jessica Wu reveals the YC playbook that got her here in 24 months. From youngest hedge fund quant to founder + the trick her batchmates missed + the Christmas Day deployment + the 10-year founder test Worth more than a year of YC advice on your timeline
Phosphen560,874 Aufrufe • vor 3 Monaten
this russian guy found a way to learn anything 10x faster🚨 saved 1,460h NotebookLM + Gemini + Obsidian dumps any source > AI strips duplicates > keeps only what u don't know yet 20 YouTube videos on the same topic each one repeats the same 20% of information this bundle removes the other 80% what used to take a month > 15 min yt: zproger
Phosphen1,043,397 Aufrufe • vor 6 Monaten
this trader made $1M on polymarket doing the most boring thing possible Anjun buys YES at 99.9¢ on markets that will obviously resolve YES makes 0.1¢ per share but buys 200,000 shares at once $202,000 position → $200 profit sounds terrible right? except he does this 10 times per day $2,000 daily $60,000 monthly $1M in under 2 years while you're trying to predict elections and find "alpha" he's just farming guaranteed 0.1-1% edges with massive size the strategy: find 99%+ certain outcomes, buy everything, wait for resolution if you have $100k sitting around doing nothing this prints $100-300 per day risk-free check him:
Phosphen1,142,331 Aufrufe • vor 7 Monaten
Dhravya Shah sold two companies before he could legally drink, then raised $3 million alone at 19. This 73-minute interview is the clearest case I've heard for why AI memory is the next platform. Here's the part I can't stop thinking about: > everyone will have their own AI agent just like everyone has their own mobile phones > in that world the most important thing will be context about you > my entire infrastructure for 100,000 users can run on $5 per month > context infrastructure for agents will be as important as inference Most builders still bolt memory onto an agent as an afterthought. The ones who get this are building the layer everything else runs on. Read how memory becomes the loop your agents run on, below.
Phosphen372,423 Aufrufe • vor 2 Monaten
DeepSeek published the models, the harness and the 88-page paper under it, then kept every researcher off camera This one talked for 59 minutes, and he explains why your agent gets dumber the longer you run it 19:04 - agents already past the self-improvement line 20:11 - why an agent wiped its folder 29:45 - why retrying makes RL models worse 41:05 - how they measured reasoning collapse 46:01 - the fix that made training faster He co-authored DeepSeek-V2 and published the training methods behind R1 Watch it before it goes the way of the founder's four-hour speech The harness guide is below
Phosphen131,953 Aufrufe • vor 1 Monat
AI bot switched 15-min to 5-min btc markets $15k today 68.7% win rate the loop (every 5 minutes): ▸ scans btc price across exchanges ▸ detects cex lag vs polymarket odds ▸ enters directional position (up or down) ▸ $1,500-$5,000 per trade ▸ exits on resolution 15-min windows = 96 trades/day 5-min windows = 288 trades/day 3x speed unlocked started with $100k rank #832 globally $124k → $1M how long?
Phosphen504,152 Aufrufe • vor 7 Monaten
This 2 hour lecture by Yann LeCun (Turing Award winner) will teach you why the next trillion dollar AI company won't be built on LLMs. He trashes the $100 Billion LLM race, attacks Musk and Amodei, declares scaling dead. Bookmark & watch tonight after work, skip to 7:00.
Phosphen261,115 Aufrufe • vor 4 Monaten
Anthropic just dropped a 3-hour course on how to actually master Claude Code, taught by the engineer who built it: 00:07 – everything new in Claude Code 24:39 – Boris Cherny builds a full app live 55:38 – the complete prompting playbook 1:29:07 – the "thinking" lever almost everyone misses 1:52:58 – how to pick the right model for the job 2:24:22 – how Anthropic itself codes with Claude This 3-hour watch will replace 10 paid Claude Code courses on the internet. Watch it now, then read how to build an AI that codes while you sleep, in the article below.
Phosphen103,234 Aufrufe • vor 2 Monaten
The two product managers who ship Grok Bot at SpaceXAI: "You should never be watching a context meter. My chat with my chief of staff bot has been running for months." In 47 minutes they put their whole bot team on a live feature and hand over the one message that builds yours. Worth more than a month of Grok Bot threads on your timeline. Watch it, then wire up your first bot with the article below.
Phosphen24,114 Aufrufe • vor 17 Tagen
Boris Cherny and Bun's creator live-coded on stage at Anthropic's conference. They shipped 3 production PRs. The audience watched. "This entire thing was one prompt, and that just ran for 30 minutes." In 32 minutes, Bun's creator walks through the setup running his repo while he sleeps. > AI bots arguing 30 rounds on every PR > a bot that out-commits Bun's creator > Bun being rewritten in Rust by Claude > the image library Claude built to beat sharp The bot writes the code. The human merges the suggestions. None of this worked three months ago.
Phosphen96,948 Aufrufe • vor 3 Monaten
He taught machines to learn. He won a Turing Award. Now he says we're building AI all wrong. "We want a path towards intelligence that isn't limited by human abilities." In 71 minutes at MIT, Richard Sutton reveals what comes after the chatbot. Self-taught + play + abstraction + superintelligence Worth more than a year of AI roadmaps on your timeline
Phosphen71,000 Aufrufe • vor 2 Monaten
Showed a friend these 50 quant formulas He dropped them into Claude and built a risk calculator for his portfolio in one prompt His first discovery: his 'profitable' strategy was actually bleeding 12% a year to variance drag and he never knew He was up 30% one month, down 30% the next, thought he broke even Actual return: -9% Two years of this. Compounding against him every single day He recalculated his entire portfolio in 10 minutes using code from the article. Found out his real Sharpe ratio was 0.3, not the 1.2 he assumed Then he ran the drawdown analysis. His max drawdown was -58% He thought it was -30% Every number he believed about his own performance was wrong These 50 concepts aren't theory. They're the exact toolkit quant desks use to see what retail never sees Full breakdown with runnable Python code below
Phosphen129,374 Aufrufe • vor 6 Monaten
SpaceXAI engineer whose Slack bot became Grok Bot, Lauren Tan (poteto): "The cat agents went wild. Refactors that would have taken a team of engineers years took two of us. I am not cooking anymore. I run a Michelin kitchen where every cook is a bot and I own the quality control." In 26 minutes she hands the bug queue to her chief of staff bot. Watch it today, then read how to set up your first Bot properly, in the article below ↓ source: MTS podcast
Phosphen13,079 Aufrufe • vor 18 Tagen
my quant friend explained how polymarket AI bots actually work btc/eth/sol/xrp 5-min markets now live they use black-scholes for binary options C_binary = e^(−rT) · N(d₂) for 5-min contracts: model says 50% probability (coin flip) but bots print hundreds of thousands how: binance confirms 0.15% move at minute 3 true probability jumps to 92% polymarket still shows 65% bot captures 27% edge before odds update the pipeline: ▸ deribit implied volatility ▸ monte carlo 200k simulations ▸ detect 3%+ model divergence ▸ execute on lag they're not predicting direction they're arbitraging information delay polymarket launching 1-min markets signal-to-noise drops 2.24x only sub-100ms systems survive when fair value updates in milliseconds who controls the orderbook? forget about losing with Poly Bot Copy ⭣
Phosphen96,336 Aufrufe • vor 7 Monaten
Best market maker on polymarket entered 5-min btc markets 0x8dxd - $1.4M profit on 15-min windows his made its name on markets now expanding to 5-min windows spread farming mechanics: YES at 48¢ + NO at 49¢ = 97¢ pair cost payout $1.00 3¢ profit locked regardless of outcome profit = min(qty_YES, qty_NO) × (1.00 − pair_cost) 3x frequency = 3x compounding speed smaller test positions for now bot adapted and kept printing Wanna outperform this AI bot? Try Poly Bot Copy One of the best copytrading bots on the market ⭣
Phosphen84,430 Aufrufe • vor 7 Monaten
NVIDIA & (ex. DeepSeek) engineer, Zihan Wang: "The model finally learned a thousand ways to say nothing It keeps very high entropy while the reasoning itself is already fragile, and the final task performance drops with it" 52 minutes on the one failure your dashboard is built to miss, the one your reward curve keeps reporting as a win Skip to 1:28, before that the host is still reading his bio Nobody uploads talks like this, they die inside a lab zoom call The harness these agents run on is broken down in the article below
Phosphen15,280 Aufrufe • vor 1 Monat
Yann LeCun just raised $1.03 billion to build what comes after LLMs. Then he spent 4 hours explaining why: "There's literally hundreds of billions invested in an industry that is basically counting on the fact that LLMs is going to reach human level intelligence. It's complete bull shit." 1:03:40 00:00 - why machine learning still sucks, in his words 57:02 - what actually comes after LLMs 1:42:34 - what he's building at AMI Labs instead 2:43:08 - JEPA, and the road to human-level AI A Turing laureate who ran Meta's AI for 12 years, backed by Bezos and NVIDIA. Bookmark this & watch it before your feed starts repeating it.
Phosphen27,249 Aufrufe • vor 2 Monaten